Data migration is a secondary feature of Komprise, which primarily moves data for archiving, but its new Elastic Data Migration capability provides a highly parallelized capability that is ideal for WAN migrations, and should increase that use case.

Krishna Subramanian, Komprise’s COO

Today, data management vendor Komprise is announcing the availability of their  Elastic Data Migration solution. It’s part of their core offering, and completely reworks its data migration capabilities by re-architecting it with a highly parallelized, multi-processing, multi-threaded approach. That moves the data much faster and less expensively than before, and is ideal in WAN heterogeneous environments, removing a hurdle to a multi-cloud strategy.

“Komprise is analytics-driven data management software, whose core use case is analyzing and archiving cold data transparently, moving it to object storage while it looks like it is in its original location,” said Krishna Subramanian, Komprise’s Chief Operating Officer. “We later added replication, and last year we added a migration capability. Customers has asked us to be able to use Komprise for migration. Point migration products like Datadobi or Data Dynamics are expensive and are one-shot products, where you just use them for that one case. So last year, we included data migration in our full product.”

The original conception of the data migration engine was to migrate data between NAS devices, within the same data centre. Increasingly, however, customers were asking to be able to do more than this.

“We have been very successful with our data migration, but increasingly, customers want to do migrations across data centres,” Subramanian said. “We designed the original data migration capability for migration within the same data centre. Latency of the WAN matters for migrations across data centres. That’s why we gutted our data migration engine. We rewrote the whole thing.”

This new Elastic Data Migration intelligently optimizes and parallelizes the workload based on available resources. This parallelism takes place at every layer of the data – shares and volumes, directories, files, and threads – adjusting to the available resources to maximize performance.

“It reduces WAN slowness by minimizing the round-trip time over the protocol during a migration needed, which eliminates the chattiness of NFS,” Subramanian said. “Rather than rely on protocols, Komprise is specifically fine-tuned to minimize overhead for each protocol. This means that customers don’t have to worry about WAN migration issues any more.

“Channel partners also asked for this to help customers with a multi-cloud strategy,” she added. “This is a really nice entry point for them.”

Komprise is supporting its assertions of enhanced efficiency with a White Paper that uses testing with the Android Benchmark to compare Komprise’s Elastic Data Migration with the data migration utility rsync. They ran multiple iterations of migrating a 74 GB Android data set using both, in an environment consisting of a source file server with 10K SAS drives and a flash-based destination file server, connected with a 10 Gbps Ethernet link, with a WAN emulator with a latency of 30 ms between the migration software and the destination file server.

“Over a LAN, we were 27 times faster than rsync, and over the WAN simulation, we were much faster than that,” Subramanian stressed. “While Komprise completed the task in minutes over the WAN, rsync had not finished after 48 hours.” Komprise was also able to cut the degradation due to higher latencies on WAN by 250%.

“Even if customers just buy us for data migration, it is much cheaper than enterprise competitors,” Subramanian said. “We are less than a third of the cost of other vendors who do this. This makes us far more competitive with the pure play data migration vendors. We are not a different market from them, particularly as we have some customers who manage over 10 PB of data with us.”

Subramanian said the new capabilities, which are included in the product with no additional cost, will significantly expand the migration use case, which translates into more sales of the complete product.

“There are customers who start with just migrations – about 20% – but they tend to buy the whole product,” she said. “They want to archive to save money but migration is one of those tactical needs. Elastic Data Management will expand this need because it handles cross-data centre migrations.”