In a world where everyone tries to spin being first, I wanted to elaborate on innovation that Maxta announced on September 9 2014. In addition to being the first and so far only, hyper-converged solution to run on the newest Intel Xeon E5 2600 v3 family, we announced Maxta as the first hyper-convergence vendor to support metrocluster for data availability and continuous availability across metro distances. Hyper-convergence is a new IT paradigm and therefore, initially, customers deployed it in non-mission critical use cases. However, as hyper-convergence gains momentum and more and more customers are becoming comfortable with the technology, there is an increasing demand to deploy it in business essential or even business critical use cases.
So why all the fuss and why is metrocluster so important? Mission critical applications entail high levels of data integrity, data availability, and continuous availability. Enterprise storage arrays have been meeting these requirements for quite some time and it is time for any hyper-convergence solution that is targeted for mission critical applications to meet these requirements as well.
High availability (HA), data availability, and data integrity are not new concepts for Maxta. We currently provide multiple capabilities to enable top-notch data integrity, data availability, and continuous availability. We are leveraging strong checksums to ensure that no bit errors can result in data corruption and our software is continuously scrubbing data to detect and correct bit errors. Like most other hyper-convergence vendors, we are leveraging synchronous local replication across servers to protect against server failures providing both data availability as well as continuous availability. While local replication can protect against storage media failures such as SSD or HDD failures too, replication is overkill for these failures. The recovery from these failures is relatively long and consumes a considerable amount of resources if replication is the only data protection feature since it requires the participation of 2 servers and consumes network bandwidth. Thus, we are also providing either RAID or mirroring protection in hardware or software to protect against media failures resulting in efficient resource utilization for recovery. The by-product of RAID/mirroring is that Maxta can support 3 SSD or disk drives failures without data loss or loss of access to data.
In some cases, failures may be more substantial. For example, a power failure may affect multiple or even all the servers in the data center. Nature-made as well as human-made disasters may affect the entire data center too. Therefore, local replication is not sufficient for comprehensive data protection. This is the reason for Maxta Metro-cluster. Maxta Metrocluster extends replication distances from few hundreds of meters to about 500 kilometers. Maxta ensures that once Metro-cluster is enabled, all data is replicated across data centers rather than across servers within a data center (technically speaking, Maxta Metro-cluster is rack-aware). Since Metrocluster is synchronous, it provides both data availability as well as continuous availability. Upon a site failure, all VMs that were running in the failed data center can be powered up immediately and resume operation at the surviving cluster.
For data availability only, replication, whether synchronous or asynchronous, requires 2 data centers and a failover process. Once it is determined that a data center failed, the failover process kicks in to recover applications at the surviving data center. This recovery may take some time especially for asynchronous replication. For continuous availability, the failover process must determine automatically that a data center failed. To avoid split brain situations, at least 3 sites are required.
The advantage and a key differentiation of Maxta as compared to any other hyper-converged solution is our support for 2 sites + tie-breaker. The tie-breaker doesn’t need to be a fully populated hyper-convergence appliance. A lightweight server running a single VM is all that is required for a tie-breaker, making Maxta Metrocluster simple and cost efficient.
Maxta is committed to maximizing the promise of hyper-convergence by providing the most choice and agility in its solutions. We also believe that there should be no compromises in the move to hyper-convergence. Our ability to innovate and be the first to support Metrocluster enables us to provide continuous availability and data availability across metro distances in a simple and cost effective way. We believe that with the addition of metro-cluster support in our portfolio along with Maxta enterprise-class data features, that there has never been a better time for businesses to hyper-converge and use software-defined storage for mission critical applications.
We welcome the opportunity to provide you with a demo or to have you trial one of our hyper-converged solutions and specifically try out high availability and data availability across distance. We are confident that you will be impressed by how advanced these solutions are and that for virtual data centers the operational benefits and savings will be significant.
– Yoram Novick, CEO & Founder, Maxta