Hyper-convergence is the consolidation of the three traditional data center tiers (compute, networking and storage) into a single converged tier. The goal of hyper-convergence is to speed time to value, allowing new infrastructure to be deployed in a fraction of the time that it previously took. Another goal is to reduce the cost of deploying these resources. Despite all the potential benefits hyper-converged infrastructures (HCI) have brought to the data center, they also have specific challenges. With its latest release (3.5) Gridstore feels it has made significant strides in overcoming these challenges.
The challenge facing potential suppliers is confusion over the term itself as vendors of all types try to jump onto the hyper-converged bandwagon. As a result there are three types of solutions that claim some form of convergence. Many solutions are the first type and are actually converged, not hyper-converged infrastructures. These solutions combine independent products into a single package. While shortening time required to stand up the infrastructure they often do little to improve the ongoing operational challenges.
The second type is software hyper-converged, solutions that run on existing server hardware and aggregate storage within those systems. These ‘bring your own’ server solutions are cost effective and, depending on their integration with the hypervisor, can also reduce operational costs. But they may actually make the time to stand up the infrastructure worse, since they require a lot of pre-integration work.
The third type, turnkey hyper-converged, provides the hardware, software, networking and storage in a single package. While potentially not as cost effective as the software-only model, they are able to provide rapid deployment and simplified ongoing integration. As we discussed in a ChalkTalk video with Gridstore founder and CTO, Kelly Murphy this is the type of solution that Gridstore provides.
Windows Integrated Hyper-Converged
An important differentiator for Gridstore is that it runs within the Microsoft Hyper-V eco-system. Maybe better said, an important differentiator for Microsoft is that Gridstore runs within its eco-system. While there are other hyper-converged architectures that can run in a Windows Hyper-V environment, most run either VMware or one of the Linux KVM variants. Let’s face it, Windows runs on the highest percentage of servers in most data centers. If you are going to convince IT professionals not to run VMware, it seems Hyper-V would be an easier path than VMware.
Another important capability of the Gridstore solution is that most of their systems are now deployed in all-flash configurations. While high capacity HDD nodes can still be added, most customers are choosing all-flash based on its simplicity, performance and affordability. Due to Gridstore’s architecture that uses Erasure Encoding instead of replication to protect against node failures – Gridstore claims to offer all-flash performance at the cost of its Hybrid Storage based competitors.
The Gridstore solution also includes quality of service (QoS) so that specific VMs can be set to specific minimum and maximum IOPS thresholds based on demand. This level of control allows for mission critical workloads to be virtualized, but to have assurance that the system will deliver user demanded performance, no matter what the load is on the rest of the environment. This also is valuable when the underlying storage fabric is all-flash. Not every workload needs this performance and some workloads need orders of magnitude higher IOPS. Gridstore’s QoS can ensure both workloads are allocated exactly what they need. This is particularly useful in cloud and mixed workload environments.
The big news in Gridstore’s 3.5 release is integration with Microsoft Azure, Microsoft’s public cloud offering. And based on the current data it’s doing well in the very competitive cloud market. Important for organizations with a bent toward Microsoft is the level of compatibility that they can enjoy with Azure. Gridstore now automates much of this integration work and can deliver turnkey cloud disaster recovery with granularity down to the VM, as well as leverage Azure for bursting or test-dev. This greatly simplifies the deployment of DR and eliminates the high entry costs of duplicating infrastructures.
Gridstore makes a compelling argument, not only for HCI but for its infrastructure specifically. The ability to cost effectively develop an all-flash HCI that provides fine grained QoS and now adds integration with the cloud for DR and bursting should broaden Gridstore’s appeal to even larger customers than before.