Customers around the globe are adopting Datrium’s Open Converged Infrastructure (OCI) platform over legacy Hyperconverged (HCI) and SAN platforms. Revenue is growing rapidly, and we’re adding hundreds of customers with a ton of successful DVX deployments. Today, I’m happy to bring you live breaking news about our newly available OCI platform, aka Datrium DVX 4.0. It has some great new features that will provide our customers (maybe you?):
- Fast retrieval of off-premises archives by restarting, not restoring applications & Cloud DVX
- ALL VMs on the host with failed SSD staying online
- Full support for Oracle RAC on a 4-node vSphere cluster
- Improved encryption manageability
- Built-in audit trail capability
- Python-based API Customer Access
If you’re not familiar with the OCI platform – unlike HCI or SAN, OCI provides simplicity, scalability and efficiency – regardless of where your data resides. Let’s take a deeper look at what the news is all about and how you might benefit.
1. Faster retrieval of off-premises archives to restart, NOT restore applications & Cloud DVX
The fast and efficient replication, storage and retrieval of VM-level data is a fundamental Datrium capability. DVX 4.0 is extending its Replication possibilities by providing a native-AWS option for users, allowing them to get their VMs off-premises to AWS for archival purposes. Unlike some competitors, Datrium Cloud DVX provides native, built-in and always-on data reduction along with forever incremental archiving to an S3 bucket. The amount of data traversing the wire is a customer concern – this is why the Datrium solution is so compelling. We’re able to minimize the amount of data exiting the data center to AWS (ingress) as well as the being retrieved back on-premises AWS (egress).
How? DVX platform global deduplication capability
Regardless of where the data originates and is targeted to, Datrium will perform a source and target cryptohash comparison of the data blocks to determine which blocks are unique versus which data already exists. The end result for a customer is bi-directional, native deduplication over the wire. In testing Cloud DVX versus S3 Direct we’ve seen a 10x reduction in AWS costs!
The initial use case for Cloud DVX is as mentioned: Archives – the data within AWS will reside in S3 making it both durable and resilient. Cloud DVX utilizes a private AMI to deploy the required EC2 instance, S3 bucket and DynamoDB all through the use of a few Datrium-written Lambda functions. These Lambda functions along with CloudWatch are also responsible for ensuring that the EC2 instance is always online.
2. ALL VMs on the host with failed SSD stay online
In the rare case where all of the SSD drives fail or the RAID controller fails in a Compute Node, Peer Cache means you’re not left with an all-paths-down (APD) situation – ALL VMs on the host with failed SSD stay online. Peer Cache builds off of Datrium’s Split Provisioning architecture, a radically different deployment methodology, as Datrium DVX now allows Compute Nodes without flash SSDs to access the flash from other Compute Nodes within the environment. Each VM is now able to leverage the flash on peer hosts within the cluster while the flashless host is repaired. In a legacy Hyperconverged world, drive failures can be tragic, leaving, in a best-case scenario, the environment in a degraded state – in a worst-case scenario recovering from backup.
How? With DVX 4.0 New Peer Cache Mode
Peer Cache Mode furthers your data availability in failure scenarios by leveraging SSD from surviving nodes within the environment. For a customer, the end result is a simple, highly resilient and scalable platform for hosting your mission critical applications and services. Datrium takes data resiliency and availability very seriously and our system is architected to ensure both high performance and resiliency. In Legacy HCI architectures, all data is persisted across each cluster node within the environment, and the default / most common deployment only allows for a single failure (RF2) before the system is in a degraded state. When a second failure occurs, applications will experience a data outage and, in some cases, data loss. Datrium ships with a default level of RF3 with Erasure Coding – meaning that 3 copies of your data are striped across all Data Nodes.
3. Fully supporting Oracle RAC
Any virtualized database is susceptible to the amount of latency introduced along with amount of I/O the disk subsystem can provide. This challenge alone makes DBAs and Administrators think twice before virtualizing. With Datrium DVX the write-path is always flash-first, providing write-intensive style applications optimal performance. In the Database use case, performance is an absolute requirement, this is why we’re requiring that customers deploy Datrium DVX, F12X2 All-Flash Data Nodes, making any read operation which is not serviced locally by the SSD drives or NVMe as fast as possible. If you experience an SSD failure, Compute Node or RAC node outage the F12X2 All-Flash Data Node will ensure that latency is as low as possible.
How? Supporting Oracle RAC on a 4-node vSphere cluster
IT is continuing to virtualize mission critical applications and services with Datrium DVX and VMware vSphere. Oracle Real Application Clusters (RAC) on VMware vSphere is a fully supported configuration, in fact this is how VMware IT runs their Oracle RAC deployment.
With the release of DVX 4.0, and the F12X2 All-Flash Data Node, the Datrium DVX platform will provide full support for Oracle RAC on a 4-node vSphere cluster making the platform even more enterprise-ready than it was before!
As proof, the Datrium Solutions team has been working with House of Brick Technologies to perform I/O benchmarking and Failover Testing of the Oracle on Datrium solution. In order to generate I/O on the Oracle Database, the benchmark tool SLOB was used. However, in official testing we wanted to mimic the behavior of an actual application, for this we utilized Swingbench. For the complete rundown on these performance tests stay tuned for Mike’s Technical Whitepaper!
4. Improved encryption Manageability
Datrium DVX 4.0 now offers the ability for organizations to set 2 decryption passphrases. In the past, if an administrator whom knew the decryption password leaves the organization, decides to reset the password to something unknown or is hit by a bus (sad panda) – there was no way to recover the passphrase. By providing the ability to set 2 passphrases the Chief Security Officer will now have a method of resetting the unknown passphrase. Think of it as a bus-factor of 2!
How? With blanket-encryption v2.0
In 2017, Datrium launched its FIPS-approved AES-XTS-256 with 52-bit key Blanket Encryption capability to providing encryption of the data in-use, over the wire and at rest. This is done with full inline compression, global deduplication and erasure coding ALWAYS-enabled all with approximately 2% overhead. This is level of data protection is something that no other converged or hyperconverged vendor is offering!
5. A built-in audit trail capability
This new auditing capability will track user actions regardless if they have authenticated through vCenter, DVX standalone UI, API or CLI. Datrium Audit Trail also monitors API actions making security analysis, change tracking, troubleshooting and auditing a breeze.
How? With a built-in native auditing feature throughout the software stack
Once you upgrade your systems to DVX 4.0 you’ll automatically be able to begin auditing actions taken in through the DVX events view. With Datrium Audit Trail your administrators will be able to log, continuously actions taken across the DVX platform.
If you’re a Healthcare organization you’re bound to comply to HIPAA, in retail you’re bound to PCI, in the publicly traded word you’re subject to Sarbanes-Oxley (SOX). All of these standards have required controls which must be audited. In my professional career, I’ve been through internal audits and external audits – none of which are fun – but they’re a necessary evil.
6. Python-based API Customer Access
Building upon the overall manageability story, Datrium DVX 4.0 will provide Python-based (initially Python only) APIs for customers to execute scripts against the DVX. With this first iteration of providing external API access, customers will be able to leverage the Python SDK to perform VM-level Get, Snapshot, Clone and Replicate tasks.
How? DVX New API Support
Datrium DVX has been written and architected with an API first approach, meaning that everything within the Datrium platform utilizes API calls behind the scenes. Up until this point we’ve consciously chosen to not expose APIs for consumption – until now.
For a full rundown on this new Python-based API visit the Downloads section of the Datrium Support site, to grab your copy of the documentation along with the full python script which is in the screenshot above.
It’s a wrap
As Datrium’s engineering efforts continue, we’re innovating at a blistering rate and building upon the many successful product releases. Don’t take my word for it, we’re so confident in DVX that we’ve decided to put our money where our mouth is. Datrium is proud to offer a customer satisfaction guarantee with every DVX system shipped which guarantees a full refund if you’re unhappy for ANY reason within 30 days after receiving the product.
I can’t express that the amount of engineering brain-power this organization has is something like I have never seen before. Kudos to the Datrium engineering and product management teams on another huge milestone!