Because data centers are built on a single-stack architecture, where separate pieces of hardware are aimed at specific applications, silos of data have been created.
When server virtualization took over, IT managers had to "retrofit 50 years of hardware architecture" to support the mixed bag of applications. "And stunningly, we have problems," Duplessie said, with a note of sarcasm.
Scale Computing, located in Indianapolis, Ind., is currently beta testing a scale-out product that converges x86 compute nodes with RedHat's KVM hypervisor and disk storage arrays.
"You can start with three nodes and add more, as needed, to increase compute or storage performance/capacity," Taneja said.
Relative to SimpliVity, Scale Computing will be targeting smaller IT shops that may or may not have server virtualization. "Hence KVM does the job very effectively for that market (lower cost, more than enough features and simplicity). Otherwise, the concepts are similar to SimpliVity," Taneja added.
OmniCube's block-level primary dedupe
OmniCube also separates itself from rivals, Kempel said, by deduplicating data at the 4K and 8K block level before it's even placed into a file format and written to the internal storage. That way, no further deduplication, compression or data rehydration is ever needed, he said. Data rehydration is the process by which deduplicated data is restored from backup appliances.
By deduplicating and compressing data up front, OmniCube also becomes a type of WAN optimization appliance, Kempel said, though it does not optimize the TCP/IP stack as other appliances from Riverbed or Silver Peak do.
VMware's VMotion data migration tool allows data and virtual machines to move between OmniCube arrays, which allows a VMware admin to increase application performance by porting VMs to OmniCubes with more SSD, Kempel said. Data on existing, third-party servers and array can also be migrated onto the OmniCubes using the same VMware tools.
The OmniCube also adds storage tiering, which takes frequently accessed data off hard drives and places it on the internal SSDs to speed up I/O performance. Additionally, the array migrates data to another OmniCube for high availability or replicates asynchronously to a remote disaster recovery site or a cloud service provider for resiliency.
The OmniCube is based on standard x86 server hardware, but the software allows data to both be served up to applications as well as held on internal capacity as primary storage.
Data replication is tied to the native backup engine in OmniCube. The policy you set for the VMs dictates when and how often you send backup copies offsite. The replication is then acting on a static backup copy, while the live VM continues in production.