Actifio pushes cloud DR that accelerates slow object storage to near SSD speed

ESG has validated Actifio’s claim it can provide public cloud disaster recovery at near SSD speed using slower, low-cost object storage in the Google Cloud Platform (GCP).

A yet-to-be-published “Microsoft SQL Server Recovery and Performance with Actifio on Google Cloud” ESG technical review paper, sponsored by Actifio, examined Actifio’s claims.

ESG used a test bed with a 967GB SQL Server production database instance running in GCP and backing it up with Actifio to SSD-class storage, which GCP calls Persistent Disk. The backup was then replicated to a different GCP region and stored there on Google Cloud Nearline (object-class) storage.

It took Actifio 90 minutes to fully back up the database and replicate it to the second DR region’s Nearline storage. A policy was set to have Actifio then back up and replicate in an incremental forever manner with application consistency. 

Then ESG initiated an on-demand Actifio Sky appliance in the DR region and used Actifio Global Manager to spin up an on-demand SQL Server host there. This host mounted the backup image in the Nearllne storage through the Sky appliance as a virtual block device.

This makes the Nearline objects look like block storage to the SQL Server host, which can only operate with block storage. Without this “spoofing” capability the Nearline objects would have to be ‘rehydrated’ to GCP’s Persistent Disk storage class.

The Actifio appliance then brought the DR SQL Server instance online, 5.5 minutes after starting the DR process. Its performance was then compared to that of the  production database.

With a 50:50 read/write mix, the Actifio DR instance running off Nearline storage, and with a Persistent Disk cache, provided 95 per cent of the performance of the production instance running on Persistent Disk. It provided 97.6 per cent of the production instance’s performance with an 80/20 read/write workload mix.


The ESG analysts then looked at the costs of this kind of DR and modelled the expected costs for a company that needed to support a 1TB SQL Server production environment over a three-year period without taking into consideration capacity and performance growth requirements or soft costs like administration. They compared the Actifio DR costs to a more traditional model of running backups and replicating full copies of the database for each DR copy. 

They stated in the paper: “The cost of storage for Actifio over three years is $76,080 compared to $648,000 in a traditional model. This represents a cost savings of $571,920, an 88 per cent cost reduction.”

Actifio’s SVP for product marketing, Chandra Reddy, said that Actifio is doing for DR what Snowflake does for data warehousing: “One of the basic characteristics of data warehousing is to grow 10s of TB to 100s of TBs or even PB very quickly and shrink back. This requires the use of a scalable storage layer. Snowflake chose to use cloud-based object storage such as AWS S3 or Google Cloud Storage or Azure Blob Storage.”

It gets faster speed than raw cloud object storage can deliver by using a scale-out design in which each “scaleout compute delivers parallel execution of parts of queries.” Also: “Each compute instance, known as a virtual warehouse (VW), caches data upon reading from object storage in its flash storage and memory. This ensures fast local data access with high IOPs for query execution.”

The on-demand Actifio Sky appliances can be scaled out too, meaning Actifio cloudDR is essentially using the same cloud model as Snowflake.