Yandex Cloud Product Digest (March 2026)

Infrastructure and network

A new availability zone — ru-central1-e — has been launched

The new availability zone is a full-fledged data center with an independent power supply and network.
Its key features are:

• The location near the ru‑central1‑a zone ensures minimal data transfer latency (less than 1 ms).
• The capacity is 40 MW, 2.8 thousand racks.
• The bandwidth is 25.6 TB/sec.
• Energy efficiency: Average PUE is 1.1 (27% lower than the global average).
• New compute platforms are available for connection, which significantly surpass the previous ones in terms of performance and vCPUs.

The new zone is perfect for hosting any new projects, including high-load applications, and the proximity to the ru‑central1‑A zone will be especially important when hosting:

• cluster applications with synchronous data replication
• high-performance databases (PostgreSQL, MySQL^®, ClickHouse^®, MongoDB)
• transactional systems and clusters with OLTP load.

You can use the new zone right now: just select ru‑central1‑e in the console when creating or migrating resources.

We wrote more about the new zone in a separate news article.

Yandex Cloud Stackland launched

A container platform that integrates Yandex Cloud PaaS services, it is deployed within an internal corporate contour and can significantly accelerate the development of applications and services and reduce time to market.

In fact, the platform allows you to deploy a whole infrastructure stack “out of the box” to develop container applications, create analytical systems, and launch AI products.

In addition to basic components like resource management tools (including GPUs) and security, Stackland includes managed databases, systems for data storage and processing, and IAM for access control.

You can deploy the platform in a few hours on your own infrastructure — virtual machines or physical servers — and start developing applications right away, allowing you to focus your resources on launching IT solutions insteading of infrastructure maintenance.

We wrote more about the launch of Yandex Cloud Stackland in a separate news article.

Dedicated server leases Yandex BareMetal

• We reduced prices for several configurations of dedicated physical servers by up to 48%. Discounted servers are available to users with the following features: 32 cores, 128-256 TB RAM, 2.6-3.3 GHz, 2 HDDs, 2-10 SSDs. You can lease these servers directly inthe console until April 17, 2026.

• We also made it possible to place an order directly in the console for an individual configuration server specifically for your needs.

• We have expanded the configurator’s capabilities — now you have a wider range of processors for ordering a server specially for your needs.

Yandex Cloud CDN News

We continue to improve Yandex Cloud CDN — today the service provides access to more than 50 points of presence. Several more new features were launched in March.

Flexible cache time management

It is now possible to granularly set the content caching time from 1 second to 1 year, allowing more accurate monitoring of data relevance.

Customizable HTTP response

This new option allows you to:

• set the redirect code from the resource,
• return a static response code with a pre-prepared response body.

This will be useful in situations where you need to force a temporary plug to a resource.

Resource blocking status, HTTP POST method, and WebSocket

The console now displays the status of resources blocks for violating the service usage rules. We remind you that using Yandex Cloud CDN to organize VPN and proxy services is forbidden .

Therefore, access to the POST HTTP method and WebSocket will be restricted (you will need to obtain an alpha flag by request via your account manager). We’ve also added control over these features to the console. Now resources allowing POST and WebSocket will be able to configure these features via the graphical console.

Following redirects in the console

The “Follow redirects from source” feature is now available right in the console interface, so there’s no need to use the API.

Data platform

Data platform services are now available in the new ru-central1-e zone:

• Yandex Managed Service for PostgreSQL
• Yandex Managed Service for MySQL^®
• Yandex Managed Service for ClickHouse^®
• Yandex MPP Analytics Engine for PostgreSQL
• Yandex Managed Service for Valkey^™
• Yandex StoreDoc
• Yandex Managed Service for Apache Airflow^®
• Yandex Managed Service for Apache Kafka^®
• Yandex Managed Service for OpenSearch
• Yandex Managed Service for Apache Spark^™
• Yandex Managed Service for Trino

We have created separate roles (managed-<cluster-type>.viewer) for viewing and editing (rescheduling) maintenance work on clusters. Now you can view and change the schedule of maintenance tasks without having access to the cluster itself.

In addition, we’ve launched user console notifications about diagnostic events related to CPU, RAM, and low memory (OOM) errors.

Aprroximate times for maintenance tasks for data platform services are now calculated, and are also included in scheduled maintenance notifications.

Yandex Managed Service for Trino news

• Support is now available for REST directory type for Iceberg^®, and for setting up connectivity to managed Apache Hive^™ Metastore clusters via Connection Manager.
• We added a Greenplum^® connector — contact the support service to start using it.
• We added support for resource group management via the API and UI.
• Private connections to a Trino cluster without access from the Internet are also available now, by sending a request to the support service.

Yandex Managed Service for Clickhouse^® news

• You can now change the disk type. To do so, you need to make sure that all tables have the Replicated engine and that the cluster is SLA-compliant. Then the disk type will change while saving all data. And, as always, we recommend making a backup in advance.
• You can now restore some databases or tables from a backup to a new cluster, making it possible to recover only the data you need more quickly, if you need to.
• You can now use self-signed certificates, e.g., from the Ministry of Finance. Please contact the support service if you are interested.
• Creating a backup via the GUI now requires additional confirmation in a pop-up window to avoid accidental clicks.
• When you have tasks to update versions, a link to the list of changes now appears, allowing you to learn about the changes in new versions:

Yandex StoreDoc news

• By default, mongos records slow requests along with full query data. This may result in sensitive information, such as personal data, being logged. We’ve added the --redactClientLogData option, which allows you to mask client request data before logging. It masks messages that accompany log events, but leaves technical metadata like error codes, line numbers, and file names unchanged. Your SIS will thank you.

• Now, if the StoreDoc replica is far behind, our automation will be able to update the information on it more quickly through backup mechanisms and avoid overloading the master.

Other news

• Yandex Managed Service for PostgreSQL: We published the support policy for PostgreSQL versions in the managed service and version update rules.
• Yandex MPP Analytics Engine for PostgreSQL: We added hybrid storage (Yezzey) and PXF extensions to the Cloudberry^™ build. Read more
• Yandex Managed Service for Apache Kafka^®: AlterConfigAdminAPI, CreateTopicAdminAPI, and DeleteTopicAdminAPI, and DeleteTopicAdminAPI events are now being delivered to Audit Trails.
• Yandex Managed Service for OpanSearch: CPU-optimized configurations are now available. You can now select cluster resources that use a higher CPU/RAM ratio.
• Yandex MetaData Hub: Support for Apache Hive^™ Metastore version 4.2.

New features in DataLens

• Background data export in XLSX. Support for saving large amounts of data from charts in XLSX format in addition to the CSV format. Background export works if the chart is a regular Wizard table, pagination and display of more than one page are enabled, the maximum number of rows in a table is 1 million, and the maximum file size is 1 GB. Read more about the restrictions in the documentation. To start background export, click Save As → XLSX in the chart menu, select All pages, wait for the pop-up notification that the export is complete, and follow the link from the message.

• Reports are even more interactive: you can now create a page with a navigation menu. Transitions between pages within a document can be customized using links: both to headings with anchors and immediately to an anchor with a page number, e.g., #page-2. It is available now and in preview modes and in PDF files exported from reports.

• In addition, Reports can now be made in a 16:9 aspect ratio and in multiple formats: FHD, 2K, 4K. It is ideal both for presentations in general and for previews on large-format screens.

How Yandex DataLens migrated from Highcharts

More than 18 million charts have been created in Yandex DataLens.
For a long time, they were all built on the Highcharts commercial visualization library. As DataLens developed, third-party dependence on a proprietary solution increasingly limited opportunities, so we got thinking about migrating.

But how do you move to a new library to visualize a product where graphics are the heart of the entire project?

Evgeny Alaev, an interface developer at Yandex DataLens, explained in an article on Habr how the team decided to write their own open-source library for visualization based on the Gravity UI design system, and how they organized seamless migration so that users would not notice anything.

Artificial intelligence

We launched the Yandex AI Studio Boost program for technology companies

Get support to build your own AI apps and agents:

• Cloud credits of up to 1 M rubles for Yandex AI Studio for six months
• Up to 70% discount for six months after receiving the grant
• Assistance with distribution and product launch strategies for partners.

The program is available to new and current Yandex Cloud customers if they have not used Yandex AI Studio in the past 180 days and have not participated in other Yandex Cloud support programs.

The program is right for you if:

• you develop and sell digital products or software, if this is part of your product or service for a foreign market,
• the exclusive rights to the software belong to you, and are not licensed,
• your company does not resell tokens.

The entire process from application to approval in Yandex AI Studio Boost can take up to 5 business days.

To get a grant:

1. Fill out the form on the site: Tell us in detail how you want to use Yandex AI Studio.
2. Have an interview: Once we receive your application, our manager will contact you to learn more about your solution and answer your questions.
3. Wait for review: After the interview, we will look at your request in detail and decide about your participation in the program.
4. Get confirmation: We send you an email about our decision, including the grant and special conditions, if you’re accepted.

New File Search features in Yandex AI Studio: search video, audio and large tables

Yandex AI Studio’s built-in tool, File Search, empowers models to generate answers based on downloaded files, rather than just the model’s general knowledge about the world.

The tool uses a RAG approach. Previously, File Search could only work with images and scans of documents, PDFs, and text. These updates will allow AI agents to search for data in tables, video, and audio.

We wrote more about File Search in a separate article. You can also read the documentation.

Video: How Yandex AI Studio pricing works

On March 3, the Yandex AI Studio platform switched to a new pricing model. In a special video, Sergey Zolotov, the platform’s technical manager, talks about how the new conditions work, as well as the new SKU line, how new tokens and tokenization work in different models, and how to calculate costs and set up spending notifications.

Video: How Yandex AI Studio security works

Along with the new pricing model, the platform has new security tools: private endpoints and managed rules for moderating model responses.

In a new video, Evgeny Baranovsky, GenAI Project Manager at Yandex AI Studio, talks about these and other aspects of information security on the platform, as well as:

• compliance with federal law, standards, and audits,
• role model and access control,
• user data isolation,
• dedicated network accesses,
• auditing and review,
• moderating model responses.

Watch on YouTube, VK Video, and Rutube.

Why tags in SpeechSense get assigned

SpeechSense has a new feature that helps understand why the system assigned a specific tag to a dialog. This increases models’ transparency and simplifies dialog analysis.

Previously, when working with tags, it wasn’t obvious why a tag had been asssigned, which meant having to read the entire dialog to understand the reason, which greatly increased the time spent on analysis.

The feature now allows you to instantly understand why a tag was triggered and reduce the time needed to analyze the dialog — you don’t need to read the entire dialog. For more information, see the relevant documentation.

Security

New version of Yandex Ruleset

Together with the SolidWall WAF team, we updated the Yandex Ruleset in Smart Web Security. We improved the coverage of current attacks and reduced the number of false positives.

Major updates:

• We’ve added 40 new rules, including 34 rules for detecting critical CVEs in common technologies and products, including Bitrix, Log4j, React, and PostgreSQL.

• We revised some of the rules from version 0.1.0 to reduce the percentage of false positives.

• We expanded the coverage of attacks like LFI, RCE, and SQLi.

The new version of Yandex Ruleset is available for selection in the WAF SWS profile.

The full list is in the description of the Yandex Ruleset ruleset.

YCDR Webinar

On March 3, we held a webinar about Yandex Cloud Detection and Response, where we discussed:

• what new cloud and on-premises event sources are available in YCDR and how monitoring coverage is expanding,
• how detection scenarios are built for hybrid infrastructure, taking Kubernetes^® and user activity into account,
• the role that Yandex Audit Trails advanced events play in incident investigations and analysis,
• how alerts are enriched with context from Yandex Security Deck and the Data Control Module (DSPM) for more accurate prioritization,
• how telemetry from different sources is combined into a single picture of an incident,
• which reduces MTTR by using centralized event correlation and ready-made response scenarios with the participation of Yandex Cloud experts.

Other news

• We took part in an AM Live broadcast about Identity Security, protecting container environments, and using ML in cybersecurity.

• We have released a detailed article about how Yandex Security Deck works.

DevOps tools

Yandex Monium

UI

• Dashboards now have the ability to set their own time interval and the default auto-update interval.

• We added the ability to display several tooltips at the same time, as well as drag and drop.

• We added support for Prometheus^® charts in notebooks, and it is now possible to customize graphs within the notebook.

• Now, in MX mode, you can add log overlays for more detailed research.

• We have expanded the dashboards copying functionality: when copying, it is now possible to automatically replace the original project with the destination project in widget requests and dashboard parameters.

Traces

• Setting up columns in search results for spam and LLM monitoring. Now, in the span search and in the LLM monitoring search, you can configure which columns will appear in the search results. When performing an analysis, you can display the most important attributes in columns and immediately see their values without going to a trace.

• Quick transition to an LLM trace from a full trace. If a trace contains spans with LLM application attributes (gen_ai.*), a button will be displayed to quickly switch to the LLM monitoring interface. This saves time in complex trace analysis — when you need to view both the spans of the AI agent itself and integrated backend systems.

• Quickly copy span attributes. Now you can quickly copy the value of any span attribute: when you hover over it, a copy button appears next to it, which places the text on the clipboard.

• Links to specific LLM Trace spans. Now you can copy and share a link not only to the entire LLM monitoring trace, but also to a specific span. As a result, you can more specifically draw the attention of colleagues to a feature when discussing your agent’s traces.

• Optimizing the performance of the LLM monitoring interface. Now the UI works faster even on large traces due to a number of data optimizations.

Alerting

We added Messenger to notification channels and escalation policies. The console has added a new type of notification and the ability to link a chat or Messenger account to a user account. Alerting has added a new channel type and a new method in escalation policies.

Other news

• Yandex Managed Service for Kubernetes^® has added support for reserve pools for cluster nodes.
• Yandex Managed Service for GitLab has added the ability to use certificates from Yandex Certificate Manager and specify custom domains for managed Gitlab instances.

Yandex Cloud Marketplace

March updates and new products in the Marketplace.

Compass

A corporate messenger for teamwork with chats, channels and threads, as well as built-in video conferencing.

The solution combines chats, calls, and collaboration in one interface: it supports group discussions, screen sharing, meeting recording, and integrations with external services. It is suitable for organizing communication within teams and working with contractors.

SecurityOnion Cluster

A solution for monitoring security and detecting threats in infrastructure.

The platform combines tools for analyzing network traffic, logs, and security events, making it possible to identify incidents, conduct investigations, and centrally manage security in the cloud.

XLTable update: OLAP server for ClickHouse^®

A solution for working with data in ClickHouse and other analytical databases. It supports the XMLA protocol, integration with LDAP, and access control. Computations are performed on the DBMS side. It is suitable for building analytics and working with data using familiar tools like Excel.

Documentation