Version 3.1 is hot off the CI/CD pipeline. This update focused on storage operators and services providers: improved controls, increased efficiency and higher availability. We also updated all of our host orchestration tools and topped the release off with a Certified Plugin for Docker Enterprise! Even better, new trial options and tutorials… Here’s the rundown of the release:
Service templates precisely define the performance characteristics of storage services. You can control how performance scales with capacity, data/network placement, and bursting policies. The software now ships with 3 predefined templates (you can create as many as you like). The ‘general purpose’ and ‘provisioned iops’ templates mimic AWS EBS. The ‘unlimited’ template mimics traditional SAN.
Blockbridge implements a rich set of tenant permissions that allow an administrator to control nearly every aspect of the tenant’s experience: from creating asynchronous replicating disks and backing up snapshots to object storage, to the types and performance levels of storage they can provision. With account templates, administrators can predefine sets of permissions for common classes of tenants. This feature was driven by a service provider who needed a way to templatize permissions based on customer type and use-case (i.e, vmware vs. openstack vs. docker vs. bare-metal).
We’ve expanded our permissions set to provide finer control over tenant functionality in service provider environments.
Configurable IO Enforcement Granularity
The dataplane now supports selectable IO sizes for quality of service, throttling and burst credits. This feature provides operator control over IOPS/bandwidth efficiency. In practice, it permits devices with low random IOPS potential to achieve high sequential bandwidth performance.
Burst credits are savings accounts for IOPS. When storage is inactive, credits are saved (up to operator defined maximum). When additional performance is required, credits are used. Burst credits are an efficiency control that drastically improves provisioning density by eliminating the need to overprovision performance relative to capacity in environments that are dominated by small volumes.
XTS Journal Bypass
Datastores configured for XTS mode now support journal bypass for sequential I/O, reducing storage bandwidth requirements and improving efficiency. When enabled, bypass is dynamic and opportunistic.
Virtual Device Automation
The disk agent microservice has been enhanced to handle storage devices that lack an identity (i.e., VMware, AWS EBS, etc). In addition, the system can be configured to automatically detect and label these devices. For example, you can now dynamically add AWS EBS devices to an instance and have it automatically configured for use in a datastore.
The disk agent now handles dynamic changes to multipath configuration and advertises the preferred paths for accessing physical storage devices. Multipath device management has been unified and simplified. We’ve also improved detection and event propagation times.
We’ve improved both our best case and worst case failover times. Worst case failure detection and service failover is now 60 seconds for a catastrophic loss of a cluster member. Failover due to common hardware and software failure modes is now as low as 10 seconds!
Hardware watchdogs are mechanisms that automatically reset system hardware in the event that software is unable to run (ex., kernel panic, kernel deadlock, dead-man switch). We’ve updated our clustering software to use watchdogs, improving availability in the event of catastrophic loss of a member.
A new tool is available to assess the configuration of your cluster and its operating system environment. The tool simplifies installation testing and reduces the probability of outages due to configuration errors.
V2 Plugin Support
A ton of stability and feature improvements went into the latest driver release including support for service templates, interface refinements, security improvements and configuration status checking. We’ve also reduced size by over 60% by switching to Alpine for the container base. The driver now supports the native docker plugin API and is now the preferred method of deployment.
Our latest volume plugin has been certified by Docker for security and functionality. The certified plugin is free to use and available from the Docker Store.
We’ve improved our customer support capabilities with new tools that allow you to securely upload diagnostics information. The diagnostics upload utility now supports client-side asymmetric encryption and HTTPS upload to a restricted access AWS S3 bucket.