
Fast speeds with Microsoft Azure Ultra Disk supporting mission-critical
applications needs a high-performance pit crew to support sustained
peak performance
Microsoft just announced the general availability of Microsoft Azure’s Ultra Disk – the fastest virtual disk in public clouds, targeted at high-performance workloads like Oracle, SQL Server, and in-memory databases like SAP HANA. It boasts 160,000 IOPS/disk with a throughput of 2,000 MBytes per second. With such fast speeds, Azure Ultra Disk is driving new dimensions of price/performance and cloud economics.
The Cost of Degraded Performance & Downtime
With super-premium high-performance characteristics (compute and storage tier), the cost of downtime increases exponentially, requiring a new paradigm for backup, near-instant restores, and cloning to minimize downtime and degradation of performance for DevOps efforts. Just like in a race car, which requires necessary pit stops, backup of databases or cloning of them, is a necessary evil. How quickly can you get going after the pit stop is critical to achieving breakthrough sustained price/performance at the application level. Else, how do you win the race?
Degraded Performance with Backup
While most legacy vendors can perform incremental forever backup for file systems and VMs, they force recurring full backups for databases, at least once a week, as shown below.

In some situations, DBAs perform recurring full and incremental dumps to disk and expect the legacy backup software to sweep from the disk. In both the solutions above, the recurring full backups lead to a large backup window and high-performance impact on production databases.
Larger the DB size, the larger the impact. You picked the faster available storage with Azure Ultra Disk to get fast application performance. The last thing you want is the pit crew (the backups) slowing you down with its recurring full backups.
Significant Downtime During Recovery
In case of a disaster, during restores, the entire database has to be restored resulting in large RTO because they have to 1) rehydrate backups (like “unzip” or “undedup”) from their proprietary dedup backup format, 2) restore, i.e. transfer data to the recovery server.
Supporting High-Performance Databases with Actifio on Azure Ultra Disk
Actifio’s Incremental Forever Approach for Super-Fast Backups and Instant Recovery
Actifio protects databases on Azure with an application-consistent, incremental forever approach using Change Block Tracking APIs. After the first full backup, Actifio backs up just the changed blocks for all subsequent backups.

This reduces the backup window and IO load on production databases by up to 20x. As shown in the picture above, after each incremental backup, it also synthesizes a point-in-time “virtual” full copy. In case of a disaster, Actifio mounts the backup image instantly, within minutes, to a recovery server in native format. Thus, the recovery time (RTO) of a 1 TB DB or a 50+ TB DB is essentially the same, i.e. just a few minutes.
Actifio has certified leading databases including SAP HANA, SQL Server and Oracle on Azure Ultra Disk and recommends both the production database and the backup copy be on Azure Ultra Disk for maximum performance. Customers can expect little or no performance degradation with Actifio’s ‘incremental forever’ approach.
You can find out more details about Actifio’s offering on Azure Marketplace.