Bulut & SaaS
Starting today, Amazon Elastic Compute Cloud (Amazon EC2) C6in instances are available in AWS Asia Pacific (Taipei). These sixth-generation network optimized instances, powered by 3rd Generation Intel Xeon Scalable processors and built on the AWS Nitro System, deliver up to 200Gbps network bandwidth, for 2x more network bandwidth over comparable fifth-generation instances. Customers can use C6in instances to scale the performance of applications such as network virtual appliances (firewalls, virtual routers, load balancers), Telco 5G User Plane Function (UPF), data analytics, high-performance computing (HPC), and CPU based AI/ML workloads. C6in instances are available in 10 different sizes with up to 128 vCPUs, including bare metal size. Amazon EC2 sixth-generation x86-based network optimized EC2 instances deliver up to 100Gbps of Amazon Elastic Block Store (Amazon EBS) bandwidth, and up to 400K IOPS. C6in instances offer Elastic Fabric Adapter (EFA) networking support on 32xlarge and metal sizes. C6in instances are available in these AWS Regions: US East (Ohio, N. Virginia), US West (N. California, Oregon), Europe (Frankfurt, Ireland, London, Milan, Paris, Spain, Stockholm, Zurich), Middle East (Bahrain, UAE), Israel (Tel Aviv), Asia Pacific (Hong Kong, Hyderabad, Jakarta, Malaysia, Melbourne, Mumbai, Osaka, Seoul, Singapore, Sydney, Taipei, Tokyo, Thailand), Africa (Cape Town), South America (Sao Paulo), Canada (Central), Canada West (Calgary), AWS GovCloud (US-West, US-East), and Mexico (Central). To learn more, visit the Amazon EC2 C6in instance page. (Kaynak dil: İngilizce.) Maestro News olayı Türkçe’ye çevirip Maestro Dev ekseninde özgün sektör analizi olarak yeniden çerçeveledi.
Bulut & SaaS
AWS, Avrupa Egemen Bulutu içinde Amazon Elastic Container Service (ECS) için gelişmiş dağıtım stratejilerinin kullanıma sunulduğunu duyurdu. Mavi/yeşil, doğrusal ve kanarya dağıtımlarını içeren bu geliştirme, kuruluşların konteynerli uygulama güncellemelerini daha fazla kontrol, azaltılmış risk ve artırılmış çeviklikle yönetmelerini sağlayarak katı veri yerleşimi ve operasyonel özerklik gereksinimleriyle uyum sağlıyor.
Bulut & SaaS
<div class="block-paragraph"><p data-block-key="2ncmz">Scientists today face challenges of extraordinary scale and complexity. From shaping and simulating the intricate dynamics of fusion plasma, to exploring the vast search space of new materials, to making sense of the exabytes of data pouring out of the world's most advanced experimental facilities. The demands on modern research are unprecedented. <a href="https://deepmind.google/research/projects/" target="_blank">Frontier AI</a> can help address these challenges, while accelerating groundbreaking scientific discoveries.</p><p data-block-key="dc2rb">In December, we shared our commitment to the White House's <a href="https://www.whitehouse.gov/presidential-actions/2025/11/launching-the-genesis-mission/" target="_blank">Genesis Mission</a> — the national effort to harness AI and double the pace of American scientific discovery within a decade. Since then, Google DeepMind (GDM) <a href="https://deepmind.google/blog/google-deepmind-supports-us-department-of-energy-on-genesis/" target="_blank">announced an early access program</a> that provides AI for science tools to all 17 Department of Energy (DOE) National Laboratories, and Google Public Sector <a href="https://cloud.google.com/blog/topics/public-sector/how-google-public-sector-and-google-deepmind-can-power-the-genesis-mission-and-a-new-era-of-scientific-discovery">shared how Gemini for Government</a> could serve as an AI backbone for the DOE.</p><p data-block-key="6uqnd">Today, at the DOE Genesis Mission Summit 2026, we are expanding this by committing $40 million of AI tokens and cloud credits for researchers in support of the Genesis Mission.</p><h3 data-block-key="bhcl0"><b>Frontier AI tools for scientific discovery</b></h3><p data-block-key="f2lp3">Under this expanded commitment, we will first provide DOE’s Genesis Mission awardees in-kind access to GDM’s frontier AI for science portfolio, including:</p><ul><li data-block-key="35aj5"><a href="https://cloud.google.com/blog/products/ai-machine-learning/alphaevolve-is-available-for-everyone"><b>AlphaEvolve</b></a> — a Gemini-powered coding and discovery agent, for designing advanced algorithms.</li><li data-block-key="dm5md"><a href="https://blog.google/innovation-and-ai/products/google-deepmind-isomorphic-alphafold-3-ai-model/" target="_blank"><b>AlphaFold 3</b></a> — a model for predicting the structure and interactions of proteins and other biomolecules.</li><li data-block-key="chpg8"><a href="https://deepmind.google/blog/alphagenome-ai-for-better-understanding-the-genome/" target="_blank"><b>AlphaGenome</b></a> — a tool for understanding how variation in DNA, including the non-coding genome, shapes biology and disease.</li><li data-block-key="294jv"><a href="https://deepmind.google/science/weathernext/" target="_blank"><b>WeatherNext</b></a> — a state-of-the-art family of AI weather forecasting models for mapping weather conditions.</li><li data-block-key="3s8k3"><a href="https://deepmind.google/blog/alphaearth-foundations-helps-map-our-planet-in-unprecedented-detail/" target="_blank"><b>AlphaEarth Foundations</b></a> — a foundational AI model for mapping and understanding our planet in unprecedented detail.</li></ul><p data-block-key="7nar0">Second, we will provide Gemini for Government seats and tokens for one year to tens of thousands of users across the DOE National Laboratories’ operations, research, and management teams. This secure platform supports the full breadth of work from the research bench to the administration of specialized user facilities serving the entire scientific community, providing a single secure foundation that the DOE mission can depend on.</p><h3 data-block-key="8bgto"><b>AI for science tools in action across the laboratory ecosystem</b></h3><p data-block-key="7s9eq">While we have a lot of work still to do, the practical impact of the Genesis Mission is already coming to life across the laboratory ecosystem.</p><p data-block-key="f6839">At <a href="https://www.pnnl.gov/" target="_blank">Pacific Northwest National Laboratory (PNNL)</a>, senior scientist Dr. Henry Kvinge is using AlphaEvolve to map out massive mathematical systems that are far too complex for humans to explore by hand. The AI uncovers hidden connections automatically, fast-tracking discoveries that would normally take researchers years to find.</p><p data-block-key="9v4f3">“Modern math relies on abstraction, but combinatorics offers concrete models that make complex geometry and algebra easier to grasp. We’ve found that systems like AlphaEvolve are perfect for this search,” said Dr. Kvinge. “By leveraging the broad mathematical knowledge of LLMs, we can automate the exploration of countless angles. We’re still experimenting, but the discoveries are already shaping our future research.”</p><p data-block-key="1hkb1">At the <a href="https://www.nlr.gov/" target="_blank">National Laboratory of the Rockies (NLR)</a>, researchers are utilizing Gemini to fundamentally change how they interact with physical laboratory hardware. Dr. Steven R. Spurgeon, a senior materials data scientist at NLR, leads a pioneering program in autonomous materials discovery.</p><p data-block-key="cklfi">"Our collaboration has allowed us to build an autonomous experimentation capability," said Dr. Spurgeon. "By deploying Gemini in our instruments, we cut microscope calibration time from over 90 minutes to about 13 minutes (eight times faster) and reduced the manual steps needed to focus an image from as many as 50 down to two. That's time and attention we've given back to the science itself, enabling genuinely autonomous workflows that observe, reason, and decide in real time. This has helped us explore parts of the material design space we simply could not have reached through manual operation alone."</p><h3 data-block-key="c0sn0"><b>Driving American innovation</b></h3><p data-block-key="eee4l">The Genesis Mission represents an opportunity to transform research and science across America. By providing access to advanced AI tools, we aim to help scientists accelerate breakthroughs across critical energy, security, and scientific challenges. To learn more about how these AI capabilities can support your research initiatives, join us at the upcoming <a href="https://events.govexec.com/google-public-sector-summit/" target="_blank">Google Public Sector Summit</a> in October.</p></div> Maestro News, gelişmeyi Maestro Dev ekseninde — yazılım, yapay zekâ, bulut ve ürün teslimatı — özgün dilde yeniden çerçeveliyor.
Bulut & SaaS
<div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">We are living in the golden age of the weekend AI side project. Thanks to agentic engineering and LLMs, the time to go from a blank IDE to a functional local application has dropped from quarters to hours. You can build your wildest ideas over a cup of coffee.</span></p> <p><span style="vertical-align: baseline;">But inside an enterprise ecosystem with rigid infrastructure and millions of users, vibe coding hits an invisible wall. Your local prototype falls apart against corporate networks, cascading errors, or getting blocked by leadership terrified of operational volatility.</span></p> <p><span style="vertical-align: baseline;">The </span><a href="https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">data</span></a><span style="vertical-align: baseline;"> is sobering: only 5% of AI prototypes make it to production; the other 95% fall into the validation abyss.</span></p> <p><span style="vertical-align: baseline;">For developers, watching people on social media ship lightning-fast AI deployments while you’re stuck in endless validation loops is maddening. To figure out how to bridge this chasm, I went into the engineering trenches at YouTube to see how they manage this exact speed-versus-risk paradox. What I discovered completely rewrites the playbook on AI software development lifecycle (SDLC) design.</span></p> <h3><span style="vertical-align: baseline;">The risk-vs-speed paradox</span></h3> <p><span style="vertical-align: baseline;">When you are solo-building, failure is cheap. Writing agentic code is like piloting a nimble jet fighter—if an AI agent misbehaves, you rewrite the prompt and instantly restart the server.</span></p> <p><span style="vertical-align: baseline;">But as AI engineering leader </span><a href="https://addyosmani.com/" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">Addy Osmani</span></a><span style="vertical-align: baseline;"> points out in our premiere of </span><a href="http://goo.gle/emergent" rel="noopener" target="_blank"><span style="font-style: italic; text-decoration: underline; vertical-align: baseline;">Emergent</span></a><span style="vertical-align: baseline;">, unconstrained agentic orchestration inside an enterprise introduces an unpredictable blast radius. Addy recalls running ten parallel agents on a personal project, context-hopping and pushing code based purely on quick previews. The technical debt accumulated fast, breaking two apps catastrophically because the modifications weren't properly isolated.</span></p> <p><span style="vertical-align: baseline;">Amplify that risk to the scale of </span><strong style="vertical-align: baseline;">YouTube</strong><span style="vertical-align: baseline;">. Its infrastructure handles billions of users on a robust, 20-year-old codebase. It is essentially a public utility; you cannot risk overloading it with experimental technical debt. Protecting a platform of this scale requires extensive, slow guardrails:</span></p></div> <div class="block-image_full_width"> <div class="article-module h-c-page"> <div class="h-c-grid"> <figure class="article-image--large h-c-grid__col h-c-grid__col--6 h-c-grid__col--offset-3 " > <img src="https://storage.googleapis.com/gweb-cloudblog-publish/images/1_Gemini_Generated_Image.max-1000x1000.jpg" alt="1_Gemini_Generated_Image"> </a> </figure> </div> </div> </div> <div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">By the time you build a primitive demo through this pipeline, the underlying AI models have evolved, leaving your idea out of date. </span><strong style="vertical-align: baseline;">How do you move at lightspeed while minimizing systemic risk? </strong></p> <h3><span style="vertical-align: baseline;">YouTube’s AI prototyping stack</span></h3> <p><span style="vertical-align: baseline;">Deepmind and former YouTube software engineer, </span><a href="https://www.linkedin.com/in/benji-bear-25972313a/" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">Benji Bear</span></a><span style="vertical-align: baseline;">, solved this puzzle not by accelerating reviews, but by changing infrastructure philosophy. He and his team built a </span><strong style="vertical-align: baseline;">prototyping stack </strong><span style="vertical-align: baseline;">— a unified design-to-code lifecycle platform that completely decouples rapid experimentation from mainline production servers. It systematically solves the two primary friction points of developer velocity.</span></p> <h3><strong style="vertical-align: baseline;">Decoupling the data layer</strong></h3> <p><span style="vertical-align: baseline;">Isolating a standalone app completely causes a "blank canvas" problem where you can't test prototypes against realistic conditions. To solve this, developers bootstrap their ideas using pre-built </span><a href="https://aistudio.google.com/" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">Google AI Studio</span></a><span style="vertical-align: baseline;"> templates. These templates hook into a proxy server set up on Google Cloud for prototype-approved read-only data. This instantly grants the prototype pre-authenticated, read-only API access to live metadata bundles (playlists, videos, channels) via strict tokens.</span></p></div> <div class="block-image_full_width"> <div class="article-module h-c-page"> <div class="h-c-grid"> <figure class="article-image--large h-c-grid__col h-c-grid__col--6 h-c-grid__col--offset-3 " > <img src="https://storage.googleapis.com/gweb-cloudblog-publish/images/2_Gemini_Generated_Image.max-1000x1000.jpg" alt="2_Gemini_Generated_Image"> </a> </figure> </div> </div> </div> <div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">Developers get the technical accuracy of live production parameters without any ability to write back to, pollute, or crash core databases. </span></p> <h3><span style="vertical-align: baseline;">Live UI injection</span></h3> <p><span style="vertical-align: baseline;">When a concept requires true real-world validation, the stack offers client-side </span><strong style="vertical-align: baseline;">YouTube Extension wrappers</strong><span style="vertical-align: baseline;">. This wrapper acts as glue code, allowing developers to inject their experimental features directly into the actual, live production web surface of YouTube.</span><span style="vertical-align: baseline;"> </span><span style="vertical-align: baseline;">Code-split chunk safeguards isolate this from production binaries, allowing prototype updates to deploy to a safe staging environment in minutes. </span></p> <p><span style="vertical-align: baseline;">The result? YouTube went from taking multiple quarters to vet an idea to launching several successful prototypes — including </span><span style="font-style: italic; vertical-align: baseline;">YouTube Recap</span><span style="vertical-align: baseline;"> and </span><span style="font-style: italic; vertical-align: baseline;">Ask YouTube </span><span style="vertical-align: baseline;">— straight to user research studies (UXR) in weeks.</span></p> <h3><span style="vertical-align: baseline;">Embrace throw-away code</span></h3> <p><span style="vertical-align: baseline;">Implementing this stack requires a profound psychological shift. Engineers are trained to treat code as permanent infrastructure, polishing and refactoring it until it’s pristine. But Benji’s core enterprise AI philosophy here is simple: </span><strong style="vertical-align: baseline;">Embrace throw-away code.</strong></p> <p><span style="vertical-align: baseline;">Google AI Studio prototypes are meant to be messy with some technical debt; their objective is to validate product-market fit using quantitative data. Trying to refactor a chaotic, AI-generated app into an enterprise codebase is an architectural trap that can create friction.</span></p></div> <div class="block-image_full_width"> <div class="article-module h-c-page"> <div class="h-c-grid"> <figure class="article-image--large h-c-grid__col h-c-grid__col--6 h-c-grid__col--offset-3 " > <img src="https://storage.googleapis.com/gweb-cloudblog-publish/images/3_Gemini_Generated_Image.max-1000x1000.jpg" alt="3_Gemini_Generated_Image"> </a> </figure> </div> </div> </div> <div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">But because Google AI Studio builds your prototype directly onto a mirrored version of production infrastructure, you establish a highly accurate baseline from day one. You still discard the messy, AI-generated script, but when an idea proves successful, rewriting it for production becomes significantly faster, cheaper, and safely positioned later in the development lifecycle—giving you a verified blueprint to code against rather than a blank canvas. </span></p> <h3><span style="vertical-align: baseline;">Move fast without breaking things</span></h3> <p><span style="vertical-align: baseline;">The core realization here is that a 95% failure rate isn’t a bug — it is the strategy. We should design environments that encourage our teams to fail more frequently and safely.</span></p> <p><span style="vertical-align: baseline;">AI has plummeted the cost of code generation. Consequently, our roles are shifting from syntax gatekeepers to </span><strong style="vertical-align: baseline;">system architects</strong><span style="vertical-align: baseline;">. Our job is to design the bridges, read-only sandboxes, and isolated pipelines that empower teams to test wild ideas without triggering catastrophic meltdowns.</span></p> <p><span style="vertical-align: baseline;">The biggest risk isn't breaking a server with messy AI code; it's missing the technological moment because validation loops are too slow. By building structural constraints that make failure safe, you give your team the freedom to run at hyper-speed.</span></p> <p><span style="font-style: italic; vertical-align: baseline;">To see the full technical breakdown, interview clips with YouTube's core infrastructure engineers, and a look inside the Google AI Studio Proto-Stack, watch our premiere episode of </span><a href="http://goo.gle/emergent" rel="noopener" target="_blank"><strong style="font-style: italic; text-decoration: underline; vertical-align: baseline;">Emergent</strong></a><span style="font-style: italic; vertical-align: baseline;"> on YouTube.</span></p></div> Maestro News, gelişmeyi Maestro Dev ekseninde — yazılım, yapay zekâ, bulut ve ürün teslimatı — özgün dilde yeniden çerçeveliyor.
Bulut & SaaS
<div class="block-paragraph_advanced"><p><a href="https://cloud.google.com/alloydb"><span style="text-decoration: underline; vertical-align: baseline;">AlloyDB</span></a><span style="vertical-align: baseline;"> is a fully managed, PostgreSQL-compatible database service built for your most demanding enterprise workloads. It combines the best of open source PostgreSQL with Google’s advanced technology, offering massive scalability, high availability, and native AI capabilities. It serves as a performant relational store, a unified backend for vector and full text search, and an analytics engine that is up to 100x faster than standard PostgreSQL. </span></p> <p><span style="vertical-align: baseline;">Vector search is the foundation of modern AI and Retrieval Augmented Generation (RAG) applications. For developers using AlloyDB and other PostgreSQL databases, </span><a href="https://github.com/pgvector/pgvector" rel="noopener" target="_blank"><code style="text-decoration: underline; vertical-align: baseline;">pgvector</code></a><span style="vertical-align: baseline;"> is a widely adopted extension for storing, indexing, and querying vector embeddings, and HNSW (Hierarchical Navigable Small World) is a highly efficient graph-based algorithm designed for approximate nearest neighbor search across multi-layered structures. With </span><a href="https://docs.cloud.google.com/alloydb/docs/ai/accelerate-with-ce"><span style="text-decoration: underline; vertical-align: baseline;">columnar engine accelerated HNSW</span></a><span style="vertical-align: baseline;"> in AlloyDB (now in preview), you can achieve up to 4x higher queries per second (QPS) for vector search compared to standard PostgreSQL HNSW.</span></p> <p><span style="vertical-align: baseline;">Enterprise AI applications face a constant trade-off between speed and accuracy. When searching through millions or billions of vectors, maximizing Queries per Second (QPS) without sacrificing search quality (recall) is critical for scaling production workloads. The PostgreSQL </span><code style="vertical-align: baseline;">pgvector</code><span style="vertical-align: baseline;"> extension offers HNSW as one of the indexes that can speed up Approximate Nearest Neighbor (ANN) searches. Let’s dive deep into how AlloyDB solves the speed vs. accuracy trade-off. </span></p> <p><span style="vertical-align: baseline;">Note: While this post focuses on HNSW performance, it’s worth noting that HNSW is just one part of AlloyDB’s advanced vector toolkit. AlloyDB also features </span><a href="https://docs.cloud.google.com/alloydb/docs/ai/create-scann-index"><span style="text-decoration: underline; vertical-align: baseline;">ScaNN</span></a><span style="vertical-align: baseline;">—a cutting-edge index backed by over 14 years of Google Research—giving you the flexibility to choose the perfect index for your workload. Additionally, for use cases demanding absolute precision, standard k-nearest neighbor (KNN) search is always available for 100% recall. Check out our </span><a href="https://docs.cloud.google.com/alloydb/docs/ai/choose-index-strategy"><span style="text-decoration: underline; vertical-align: baseline;">Choose a Vector Index Guide</span></a><span style="vertical-align: baseline;"> to see how they stack up.</span></p></div> <div class="block-aside"><dl> <dt>aside_block</dt> <dd>&lt;ListValue: [StructValue([(&#x27;title&#x27;, &#x27;Get started with a 30-day AlloyDB free trial instance&#x27;), (&#x27;body&#x27;, &lt;wagtail.rich_text.RichText object at 0x7f019660e400&gt;), (&#x27;btn_text&#x27;, &#x27;Start building for free&#x27;), (&#x27;href&#x27;, &#x27;http://goo.gle/try_alloydb&#x27;), (&#x27;image&#x27;, None)])]&gt;</dd> </dl></div> <div class="block-paragraph_advanced"><h3><strong style="vertical-align: baseline;">First, what is the AlloyDB columnar engine? </strong></h3> <p><span style="vertical-align: baseline;">The </span><a href="https://docs.cloud.google.com/alloydb/docs/columnar-engine/about"><span style="text-decoration: underline; vertical-align: baseline;">AlloyDB columnar engine</span></a><span style="vertical-align: baseline;"> is a built-in, in-memory cache that automatically stores frequently queried data in a specialized, scan-optimized columnar format. It allows AlloyDB to handle heavy analytical queries up to 100x faster than standard PostgreSQL. Additionally, it accelerates ANN searches by storing the index in memory, using a vectorized memory layout for fast traversals, and bypassing standard PostgreSQL buffer manager overhead. </span></p> <h3><strong style="vertical-align: baseline;">Performance visualization</strong></h3> <p><span style="vertical-align: baseline;">To understand the real-world performance characteristics of columnar engine Accelerated HNSW, we plotted standard QPS vs Recall curves for the GloVe 100 Angular dataset by searching more than 1M records with a limit of 100.</span></p> <p><span style="vertical-align: baseline;">Running this </span><a href="https://colab.research.google.com/github/GoogleCloudPlatform/python-docs-samples/blob/main/alloydb/notebooks/columnar_engine_accelerated_hnsw_vector_search_benchmark.ipynb" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">benchmark script</span></a><span style="vertical-align: baseline;"> yields the following visualization:</span></p></div> <div class="block-image_full_width"> <div class="article-module h-c-page"> <div class="h-c-grid"> <figure class="article-image--large h-c-grid__col h-c-grid__col--6 h-c-grid__col--offset-3 " > <img src="https://storage.googleapis.com/gweb-cloudblog-publish/images/1_r57mjyN.max-1000x1000.png" alt="1"> </a> <figcaption class="article-image__caption "><p data-block-key="xai0q">Note: These measurements were taken on an AlloyDB C4A 16vCPU machine. Due to the inherent randomness in HNSW graph building, results may slightly vary across runs.</p></figcaption> </figure> </div> </div> </div> <div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">The data reveals two transformative benefits:</span></p> <ol> <li role="presentation"><strong style="vertical-align: baseline;">Massive performance throughput gains: </strong><span style="vertical-align: baseline;">For any given target recall (e.g. 0.95), QPS is increased by approximately </span><strong style="vertical-align: baseline;">4.2x to 4.9x</strong><span style="vertical-align: baseline;">. This allows you to handle significantly more concurrent vector searches on the same hardware.</span></li> <li role="presentation"><strong style="vertical-align: baseline;">Significant recall (accuracy) improvement: </strong><span style="vertical-align: baseline;">Conversely, at a fixed QPS level, columnar engine accelerated HNSW provides a substantial boost in recall. For example, we saw that at ~350 QPS (in the above chart), enabling the columnar engine improves recall from roughly </span><strong style="vertical-align: baseline;">0.78 to over 0.94 </strong><span style="vertical-align: baseline;">– a </span><strong style="vertical-align: baseline;">0.163 recall gain</strong><span style="vertical-align: baseline;">. This means your AI applications get much more accurate results without any latency impact.</span></li> </ol> <p><span style="vertical-align: baseline;">It is important to note that the baseline (blue line) already represents the index being fully cached in the PostgreSQL shared buffer cache. The performance gains shown here are not the result of moving data from disk to RAM, but rather the result of a more efficient memory architecture.</span></p> <h3><strong style="vertical-align: baseline;">How it works: Columnar engine Accelerated HNSW</strong></h3> <p><span style="vertical-align: baseline;">In standard PostgreSQL architectures, index operations utilize the shared buffer cache. Even when data is fully in-memory, the database still incurs significant overhead from the buffer manager, which must handle operations such as page pinning and unpinning, lock acquisition, buffer table lookups, and Least Recently Used (LRU) management.</span></p> <p><span style="vertical-align: baseline;">AlloyDB's </span><strong style="vertical-align: baseline;">columnar engine </strong><span style="vertical-align: baseline;">is a built-in, in-memory cache that stores data in a specialized, scan-optimized format.</span></p> <p><span style="vertical-align: baseline;">With this release, AlloyDB can use </span><strong style="vertical-align: baseline;">columnar engine accelerated HNSW </strong><span style="vertical-align: baseline;">to:</span></p> <ul> <li role="presentation"><strong style="vertical-align: baseline;">Pin the index: </strong><span style="vertical-align: baseline;">The </span><code style="vertical-align: baseline;">pgvector</code><span style="vertical-align: baseline;"> HNSW index is pinned (kept persistently in-memory to ensure fast access) directly into the columnar engine’s memory.</span></li> <li role="presentation"><strong style="vertical-align: baseline;">Vectorized access: </strong><span style="vertical-align: baseline;">It utilizes a memory layout specifically designed for the high-concurrency, pointer-heavy traversals required by HNSW graphs.</span></li> <li role="presentation"><strong style="vertical-align: baseline;">Bypass buffer overhead: </strong><span style="vertical-align: baseline;">By navigating the graph in a specialized memory space, AlloyDB avoids the standard buffer manager bottlenecks. This architectural shift is what enables the dramatic QPS and recall improvements shown above, even when comparing against a fully-cached standard index.</span></li> </ul> <h3><strong style="vertical-align: baseline;">Why it Matters</strong></h3> <p><span style="vertical-align: baseline;">For enterprise-scale applications, this isn't just about a faster database—it's about cost and quality:</span></p> <ul> <li aria-level="1" style="list-style-type: disc; vertical-align: baseline;"> <p role="presentation"><strong style="vertical-align: baseline;">Reduced infrastructure costs:</strong><span style="vertical-align: baseline;"> Achieve the same performance with significantly lower compute resources.</span></p> </li> <li aria-level="1" style="list-style-type: disc; vertical-align: baseline;"> <p role="presentation"><strong style="vertical-align: baseline;">Better AI accuracy:</strong><span style="vertical-align: baseline;"> Reach higher recall and quality at speeds that were previously only possible for "draft" (high-speed, lower-accuracy results) quality search.</span></p> </li> <li aria-level="1" style="list-style-type: disc; vertical-align: baseline;"> <p role="presentation"><strong style="vertical-align: baseline;">No application changes required:</strong><span style="vertical-align: baseline;"> Because this is built into AlloyDB, you get these gains using the same standard </span><code style="vertical-align: baseline;">pgvector</code><span style="vertical-align: baseline;"> SQL syntax.</span></p> </li> </ul> <p><span style="vertical-align: baseline;">Note that the columnar engine does utilize memory, but it is highly compressed and meticulously managed. Because the engine stores vector data in an efficient columnar format, the memory footprint is minimal compared to the massive performance gains—making it a highly favorable trade-off for enterprise workloads.</span></p> <h3><strong style="vertical-align: baseline;">Quick Start Guide</strong></h3> <p><span style="vertical-align: baseline;">To try out </span><a href="https://docs.cloud.google.com/alloydb/docs/ai/accelerate-with-ce"><strong style="text-decoration: underline; vertical-align: baseline;">columnar engine accelerated HNSW</strong></a><span style="vertical-align: baseline;"> in AlloyDB, follow these steps:</span></p> <p><span style="vertical-align: baseline;">1. </span><strong><a href="https://docs.cloud.google.com/alloydb/docs/ai/accelerate-with-ce"><span style="text-decoration: underline; vertical-align: baseline;">Enable the columnar engine</span></a><span style="vertical-align: baseline;"> and index caching</span></strong></p> <p><span style="vertical-align: baseline;">Ensure that both </span><code style="vertical-align: baseline;">google_columnar_engine.enabled</code><span style="vertical-align: baseline;"> and </span><code style="vertical-align: baseline;">google_columnar_engine.enable_index_caching</code><span style="vertical-align: baseline;"> flags are set to </span><code style="vertical-align: baseline;">on</code><span style="vertical-align: baseline;"> for your AlloyDB instance.</span></p> <p><span style="vertical-align: baseline;">2. <strong>Add the HNSW Index to columnar engine</strong></span></p> <p><span style="vertical-align: baseline;">Once your HNSW index is created via </span><code style="vertical-align: baseline;">pgvector</code><span style="vertical-align: baseline;">, execute the following SQL command to cache it in the columnar engine:</span></p></div> <div class="block-code"><dl> <dt>code_block</dt> <dd>&lt;ListValue: [StructValue([(&#x27;code&#x27;, &quot;SELECT google_columnar_engine_add_index(&#x27;&lt;hnsw_index_name&gt;&#x27;);&quot;), (&#x27;language&#x27;, &#x27;&#x27;), (&#x27;caption&#x27;, &lt;wagtail.rich_text.RichText object at 0x7f0196625d90&gt;)])]&gt;</dd> </dl></div> <div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">3. <strong>Additional Resources</strong></span></p> <ul> <li aria-level="1" style="list-style-type: disc; vertical-align: baseline;"> <p role="presentation"><span style="vertical-align: baseline;">New to AlloyDB? Discover AlloyDB with a </span><a href="https://docs.cloud.google.com/alloydb/docs/free-trial-cluster"><span style="text-decoration: underline; vertical-align: baseline;">30-day free trial</span></a><span style="vertical-align: baseline;">.</span></p> </li> <li aria-level="1" style="list-style-type: disc; vertical-align: baseline;"> <p role="presentation"><a href="https://colab.research.google.com/github/GoogleCloudPlatform/python-docs-samples/blob/main/alloydb/notebooks/columnar_engine_accelerated_hnsw_vector_search_benchmark.ipynb" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">Google Colab Notebook</span></a><span style="vertical-align: baseline;">: An end-to-end Python script to ingest the GloVe dataset, create indexes, and plot Recall vs QPS curves.</span></p> </li> <li aria-level="1" style="list-style-type: disc; vertical-align: baseline;"> <p role="presentation"><span style="vertical-align: baseline;">Is HNSW the right vector index choice for your use case? Check our ‘</span><a href="https://docs.cloud.google.com/alloydb/docs/ai/choose-index-strategy"><span style="text-decoration: underline; vertical-align: baseline;">Choose a vector index in AlloyDB AI</span></a><span style="vertical-align: baseline;">’ guide.</span></p> </li> </ul></div> Maestro News, gelişmeyi Maestro Dev ekseninde — yazılım, yapay zekâ, bulut ve ürün teslimatı — özgün dilde yeniden çerçeveliyor.
Bulut & SaaS
<div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">As adversarial AI threats accelerate attacks on code, security teams must counter them with machine-speed defenses that can automate code remediation and fight AI with AI.</span></p> <p><a href="https://cloud.google.com/security/codemender">CodeMender</a> is our managed code security agent, and starting today, we're bringing its code scanning and remediation capabilities directly to you in preview.</p> <p><span style="vertical-align: baseline;">CodeMender offers access to our generally available models via </span><a href="https://docs.cloud.google.com/gemini-enterprise-agent-platform/codemender"><span style="text-decoration: underline; vertical-align: baseline;">Gemini Enterprise Agent Platform</span></a><span style="vertical-align: baseline;">, or it can be deployed as a core component of </span><a href="https://cloud.google.com/security/ai-threat-defense"><span style="text-decoration: underline; vertical-align: baseline;">AI Threat Defense</span></a><span style="vertical-align: baseline;">. </span></p> <p><span style="vertical-align: baseline;">CodeMender also aligns with our </span><a href="https://cloud.google.com/blog/products/identity-security/cloud-ciso-perspectives-next-26-why-we-re-multicloud-and-multi-ai"><span style="text-decoration: underline; vertical-align: baseline;">multi-model approach</span></a><span style="vertical-align: baseline;">, so you can choose the right model to optimize for cost, speed, and deep scanning performance. It will support third-party frontier model options later this year.</span></p></div> <div class="block-video"> <div class="article-module article-video "> <figure> <a class="h-c-video h-c-video--marquee" href="https://youtube.com/watch?v=4DJD3RHOnPA" data-glue-modal-trigger="uni-modal-4DJD3RHOnPA-" data-glue-modal-disabled-on-mobile="true"> <div class="article-video__aspect-image" style="background-image: url(https://storage.googleapis.com/gweb-cloudblog-publish/images/1_sg64BeM.max-1000x1000.png);"> <span class="h-u-visually-hidden">How to find and fix code vulnerabilities autonomously with Google CodeMender.</span> </div> <svg role="img" class="h-c-video__play h-c-icon h-c-icon--color-white"> <use xlink:href="#mi-youtube-icon"></use> </svg> </a> <figcaption class="article-video__caption h-c-page"> <h4 class="h-c-headline h-c-headline--four h-u-font-weight-medium h-u-mt-std">Watch this overview of CodeMender in Gemini Enterprise Agent Platform.</h4> </figcaption> </figure> </div> <div class="h-c-modal--video" data-glue-modal="uni-modal-4DJD3RHOnPA-" data-glue-modal-close-label="Close Dialog"> <a class="glue-yt-video" data-glue-yt-video-autoplay="true" data-glue-yt-video-height="99%" data-glue-yt-video-vid="4DJD3RHOnPA" data-glue-yt-video-width="100%" href="https://youtube.com/watch?v=4DJD3RHOnPA" ng-cloak> </a> </div> </div> <div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">CodeMender can help you advance from passive scanning to automated code remediation, and reduce zero-day risk. It examines and remediates existing code security issues without sacrificing development velocity by:</span></p> <ul> <li aria-level="1" style="list-style-type: disc; vertical-align: baseline;"> <p role="presentation"><strong style="vertical-align: baseline;">Deploying the best-fit model</strong><span style="vertical-align: baseline;">. You can choose from multiple models to optimize for costs, speed, deep scanning, and coding performance.</span></p> </li> <li aria-level="1" style="list-style-type: disc; vertical-align: baseline;"> <p role="presentation"><strong style="vertical-align: baseline;">Automating machine-scale remediation</strong><span style="vertical-align: baseline;">. You can now eliminate remediation bottlenecks caused by manual verification and patching, while keeping developers in the loop.</span></p> </li> <li aria-level="1" style="list-style-type: disc; vertical-align: baseline;"> <p role="presentation"><strong style="vertical-align: baseline;">Prioritizing fixes by exploitability</strong><span style="vertical-align: baseline;">. You can run proof-of-concept exploits and execute simulations to verify that vulnerabilities in the code are exploitable, and prioritize resources on fixing the most critical issues first.</span></p> </li> </ul> <h3><strong style="vertical-align: baseline;">Find and fix vulnerabilities with AI</strong></h3> <p><span style="vertical-align: baseline;">Born from </span><a href="https://deepmind.google/blog/introducing-gemini-3-5-flash-cyber" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">Google DeepMind's pioneering AI research</span></a><span style="vertical-align: baseline;">, CodeMender transforms vulnerability management from a manual bottleneck into an autonomous, high-speed system. Your developers and security practitioners can automatically scan software for flaws, verify them with executable exploits, and remediate them with tested code fixes. </span></p> <p><span style="vertical-align: baseline;">“At Salesforce, trust is our number one value, and protecting customer data means continually raising the bar for how we find, validate, and mitigate risks. CodeMender brings AI into a critical part of the security lifecycle by accelerating the path from validated vulnerability to tested fix. As AI reshapes the threat landscape, capabilities like this help strengthen resilience and give our customers the confidence to keep innovating,” said Iain </span><span style="vertical-align: baseline;">Mulholland, CISO, Salesforce</span><span style="vertical-align: baseline;">.</span></p> <p><span style="vertical-align: baseline;">"CodeMender consistently identified critical vulnerabilities that our other AI-enabled tools completely missed. It doesn't just find theoretical flaws — it proves the immediate risk and delivers targeted, validated fixes that secure our environment without disrupting core business logic," said Scott Ponte, head, Security Operations, Robinhood. </span></p> <p><span style="vertical-align: baseline;">"CodeMender is fast, comprehensive, and genuinely ambitious about closing the loop from detection to fix, enabling teams to secure their software supply chain without losing velocity," said Ashwin Kannan, principal AI engineer, Office of the CTO, Palo Alto Networks.</span></p> <h3><strong style="vertical-align: baseline;">How the CodeMender agent works</strong></h3> <p><span style="vertical-align: baseline;">We’ve fine-tuned CodeMender’s harness to be continuously updated with the latest Google DeepMind research, including the up-to-date agent skills, security tools, and system prompts. </span></p> <p><span style="vertical-align: baseline;">Operating in the secure-by-design Agent Platform, CodeMender is protected by enterprise-grade, built-in governance and security guardrails, including secure traffic routing through your VPC, data isolation and encryption, and zero retention of source code data.</span></p> <p><span style="vertical-align: baseline;">As an agent, it can integrate with existing continuous integration and continuous delivery (CI/CD) workflows, or run directly in local developer environments using a lightweight command-line interface (CLI) client. </span></p> <p><span style="vertical-align: baseline;">You can also configure CodeMender to scan and analyze code in a sandbox that you manage. The agent connects to your code repositories and works with developer tools, such as </span><a href="https://docs.cloud.google.com/code/docs/vscode/install"><span style="text-decoration: underline; vertical-align: baseline;">VS Code</span></a><span style="vertical-align: baseline;"> and </span><a href="https://antigravity.google/" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">Antigravity</span></a><span style="vertical-align: baseline;">, to safely analyze first-party, open-source, and third-party software.</span></p> <h3><strong style="vertical-align: baseline;">Scan: Find hidden vulnerabilities with flexible model scanning </strong></h3> <p><span style="vertical-align: baseline;">CodeMender scans for top vulnerability classes and understands the </span><a href="https://cloud.google.com/blog/products/identity-security/cloud-ciso-perspectives-how-ai-leverages-deep-context-defenders-advantage"><span style="text-decoration: underline; vertical-align: baseline;">unique context, goals, and functionality</span></a><span style="vertical-align: baseline;"> of your software repositories and applications.</span></p></div> <div class="block-image_full_width"> <div class="article-module h-c-page"> <div class="h-c-grid"> <figure class="article-image--large h-c-grid__col h-c-grid__col--6 h-c-grid__col--offset-3 " > <img src="https://storage.googleapis.com/gweb-cloudblog-publish/images/2_LNSezkk.max-1000x1000.png" alt="2"> </a> <figcaption class="article-image__caption "><p data-block-key="c7u8w">Scan: Discovered new vulnerabilities and categorized by severity and type.</p></figcaption> </figure> </div> </div> </div> <div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">CodeMender’s harness with security context helps you discover sophisticated vulnerabilities that static and model-only scanning miss. These scans look for hard-to-find vulnerabilities like memory corruption, injection, web security issues, cryptographic flaws, and insecure data handling. CodeMender supports common software languages including C/C++, Go, Java, Python, Ruby, Rust, and TypeScript.</span></p> <h3><strong style="vertical-align: baseline;">Verify: Simulate and verify exploits to reduce noise</strong></h3> <p><span style="vertical-align: baseline;">CodeMender can help cut alert fatigue and false positives by proving a vulnerability presents a legitimate risk before fixing it. The agent goes beyond static code-pattern analysis by simulating an attack with exploit code it builds and runs in an isolated, customer-managed sandbox.</span></p></div> <div class="block-image_full_width"> <div class="article-module h-c-page"> <div class="h-c-grid"> <figure class="article-image--large h-c-grid__col h-c-grid__col--6 h-c-grid__col--offset-3 " > <img src="https://storage.googleapis.com/gweb-cloudblog-publish/images/3_OEvpSjA.max-1000x1000.png" alt="3"> </a> <figcaption class="article-image__caption "><p data-block-key="c7u8w">Verify: Creates verification plan and builds and tests exploits in your sandbox environment.</p></figcaption> </figure> </div> </div> </div> <div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">The agent uses this proof-of-concept exploit to verify that the security flaw poses a legitimate risk. This critical verification phase allows your security practitioners and developers to prioritize validated risks by eliminating false positives.</span></p> <h3><strong style="vertical-align: baseline;">Remediate: Automatically generate and test code fixes</strong></h3> <p><span style="vertical-align: baseline;">Identifying risky security flaws is only half the battle. Once a vulnerability is verified, CodeMender automatically generates a secure patch to resolve the issue. The fix is delivered as a code difference directly in developer tools, so it can be integrated into existing development workflows.</span></p></div> <div class="block-image_full_width"> <div class="article-module h-c-page"> <div class="h-c-grid"> <figure class="article-image--large h-c-grid__col h-c-grid__col--6 h-c-grid__col--offset-3 " > <img src="https://storage.googleapis.com/gweb-cloudblog-publish/images/4_LlL5FCA.max-1000x1000.png" alt="4"> </a> <figcaption class="article-image__caption "><p data-block-key="c7u8w">Remediate: Generates and tests code fix with code diff for developer review and approval.</p></figcaption> </figure> </div> </div> </div> <div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">CodeMender further strengthens the fix by using LLM-as-a-judge to ensure it doesn’t disrupt existing application functionality. You can even provide context on your codebase's distinct coding conventions and styles so that CodeMender generates code that matches it. Developers remain in full control, manually reviewing and approving CodeMender's patches before any code is committed to the repository.</span></p> <h3><strong style="vertical-align: baseline;">CodeMender in AI Threat Defense</strong></h3> <p><span style="vertical-align: baseline;">When leveraged as part of </span><a href="https://cloud.google.com/security/ai-threat-defense"><span style="text-decoration: underline; vertical-align: baseline;">AI Threat Defense</span></a><span style="vertical-align: baseline;">, Wiz orchestrates agentic application security, analyzing applications to prioritize investigations. It calls CodeMender to scan code (coming soon), enrich findings within the </span><a href="https://www.wiz.io/lp/wiz-security-graph" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">Wiz Security Graph</span></a><span style="vertical-align: baseline;"> with deployment context, and trigger </span><a href="https://www.wiz.io/solutions/red-agent" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">Wiz Red Agent</span></a><span style="vertical-align: baseline;"> for AI pentesting to prove exploitability, ensuring that teams focus on the highest-risk vulnerabilities.</span></p></div> <div class="block-image_full_width"> <div class="article-module h-c-page"> <div class="h-c-grid"> <figure class="article-image--large h-c-grid__col h-c-grid__col--6 h-c-grid__col--offset-3 " > <img src="https://storage.googleapis.com/gweb-cloudblog-publish/original_images/AITD_Wheel_-_Copy_of_Final_-_BLOG-ALT_AIThreatChart_2436x1200_v2.gif" alt="AITD Wheel - Copy of Final - BLOG-ALT_AIThreatChart_2436x1200_v2"> </a> <figcaption class="article-image__caption "><p data-block-key="r3bx6">Through Wiz, AI Threat Defense calls CodeMender to scan code, enrich findings, and trigger AI pentesting.</p></figcaption> </figure> </div> </div> </div> <div class="block-paragraph_advanced"><p><span style="vertical-align: baseline;">Wiz serves as a command center for governing and scaling remediation in AI Threat Defense. The </span><a href="https://www.wiz.io/blog/introducing-wiz-green-agent" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">Wiz Green Agent</span></a><span style="vertical-align: baseline;"> orchestrates this lifecycle by directing CodeMender to generate and test high-fidelity patches enriched with application context from the Security Graph. This </span><a href="https://www.wiz.io/blog/introducing-wiz-workflows" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">workflow</span></a><span style="vertical-align: baseline;"> empowers teams to resolve complex vulnerabilities with unprecedented speed and precision.</span></p> <h3><strong style="vertical-align: baseline;">How to get started with CodeMender</strong></h3> <p><span style="vertical-align: baseline;">Consistent with our </span><a href="https://cloud.google.com/blog/products/identity-security/cloud-ciso-perspectives-next-26-why-we-re-multicloud-and-multi-ai"><span style="text-decoration: underline; vertical-align: baseline;">multi-model approach</span></a><span style="vertical-align: baseline;">, CodeMender can help you optimize for cost, speed, and deep scanning performance.</span></p> <p><span style="vertical-align: baseline;">You can use CodeMender with our generally available Gemini models via </span><a href="https://docs.cloud.google.com/gemini-enterprise-agent-platform/codemender"><span style="text-decoration: underline; vertical-align: baseline;">Agent Platform</span></a><span style="vertical-align: baseline;">, or deploy it as a core component of </span><a href="https://cloud.google.com/security/ai-threat-defense"><span style="text-decoration: underline; vertical-align: baseline;">AI Threat Defense</span></a><span style="vertical-align: baseline;">.</span></p> <p><span style="vertical-align: baseline;">Separately, CodeMender with </span><a href="https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-6-flash-3-5-flash-lite-3-5-flash-cyber/" rel="noopener" target="_blank"><span style="text-decoration: underline; vertical-align: baseline;">Gemini 3.5 Flash Cyber</span></a><span style="vertical-align: baseline;"> will be exclusively available to a small set of governments and trusted partners. We plan to expand this access over time.</span></p> <p><span style="vertical-align: baseline;">CodeMender is a critical step towards a continuous, self-healing agentic software development lifecycle, a future where code is autonomously secured, validated, and patched before it ever hits production. </span></p> <p><span style="vertical-align: baseline;">You can learn more about CodeMender and review the documentation </span><a href="https://docs.cloud.google.com/gemini-enterprise-agent-platform/codemender"><span style="text-decoration: underline; vertical-align: baseline;">here</span></a><span style="vertical-align: baseline;">. </span></p></div> Maestro News, gelişmeyi Maestro Dev ekseninde — yazılım, yapay zekâ, bulut ve ürün teslimatı — özgün dilde yeniden çerçeveliyor.
