HyperConverged Infrastructure (HCI) is a hot topic these days. It’s promise is increased flexibility (agility) and scalability at a lower cost than infrastructures based on traditional shared storage (SAN/NAS) while retaining the functionality that we have become accustomed to (like snapshots, cloning and replication) and supporting most enterprise application environments.
Creating a hyperconverged infrastructure can be achieved very easily with the building block concept of companies like Nutanix and Simplivity. You can start very simple by creating a cluster of (usually) only 3 nodes and gradually increase the size of your infrastructure when required by simply adding nodes to the existing environment. The potential disadvantage of this (hardware) appliance-based approach is that you are usually limited in your choice of appliance configuration (amount of CPU, memory and type/capacity of storage devices). This means that you could end up with an infrastructure with too much CPU, too much memory or too much storage capacity (so too much investment …).
A second approach, instead of buying new hardware, is using your existing server (brand) with a software-only HCI solution. This way you can configure the infrastructure to meet your exact specific resource needs and scale only those resource components (either CPU, memory or disk) as and when required. One such software-based solution is VMware Virtual SAN. Another one I recently came across is from a company called Springpath.
Springpath came out of stealth earlier this year after nearly three years of development by former VMware engineering talent who were previously involved in developing VMware storage and networking functionality. The product that was launched is called the Springpath Data Platform and is based on their proprietary architecture called HALO (Hardware Agnostic Log-structured Objects).
The solution uses an existing server environment with local disks (combination of SSD and HDD) to create a shared storage resource pool accessible to all servers in the cluster. By virtue of not having any storage legacy the solution has all the characteristics that fits a modern storage solution like using SSD (flash) for performance and HDD for capacity (without “traditional tiering”) and data reduction through inline compression and de-duplication. Like many other modern storage solutions it is based on a very sophisticated file system which amongst others allows for a very efficient snapshot and cloning mechanism. This can be a great benefit for VDI environments (creating hundreds of virtual desktops in minutes) or for DevOps.
Management of the solution is done through a very simple interface integrated into the hypervisor management software. Combined with the fact that you can use your own (well known) brand of server supplier there is no need to learn new interfaces making management a lot easier (usually also resulting in less errors and higher productivity).
Although there are several similarities between VMware Virtual SAN and Springpath Data Platform the latter distinguishes itself by being able to report health and usage parameters to the Springpath cloud environment allowing for pro-active support by the vendor. Also, although currently the Springpath product supports a VMware vSphere environment, it is reported that in the future the solution will support other hypervisors.
Springpath licensing is quite simple and unique for a “storage solution” : an annual subscription per node (not CPU) which allows for a “pay-as-you-grow” model without high upfront investment.
I’m looking forward to check out this solution in the near future and see how this compares to the likes of VMware Virtual SAN. To be continued …