Sometimes a technology captures your imagination and shows real promise for developing into something of real value for the customer. Scale Computing is one such company that has continued to demonstrate technical innovation and a continuning evolution in their platform. The most recent announcements bring to the market higher performance and capacity appliances, including an all-flash system. Whilst these products may not be first to the market in these categories, there’s a clear path to move from SMB to more of an enterprise focus.
I first saw the detail of Scale Computing’s hyper-converged architecture at at Tech Field Day event back in 2014. From a purely technical standpoint, features like SCRIBE and the State Engine were fascinating learn about and understand in the context of how hyper-converged was developing at the time. Remember that in April 2014, Nutanix and Simplivity were the big dogs in HCI, with VMware’s Virtual SAN having just been released the month prior. Where Nutanix and Simplivity used VMware vSphere with virtual machines for running the storage layer, Scale Computing took a different route with their HC3 platform and built an entire ecosystem based on KVM. This meant building out a distributed file system (SCRIBE) and tools to monitor and manage the hardware (the state engine).
Rather than target enterprise customers (and go up against the incumbents), Scale aimed at the SMB, with smaller, cheaper entry-level solutions, including one I have in the lab today. So far this approach seems to have been a positive strategy, with an increase in customers from 1500 when I last wrote about the company in May 2016, to 2500 today. As the customer base has increased, Scale has released more powerful systems, with hybrid (flash/disk) systems introduced in May 2016 and in July 2017 the first all-flash system.
The new HC1150DF has dual Intel Xeon E5-2620v4 processors, an entry point of 128GB of DRAM, with 4x 960GB SSDs. Nodes can be upgraded with up to 512GB of DRAM and 4x 1.92TB drives. Compared to previous systems, this is a sizeable jump in specification, especially when put in context against the other hardware platform released at the same time – the new HC5150D. This is a dual processor system with up to 768GB of DRAM and a hybrid storage configuration of 9x 8TB HDD and 3x 1.92TB SSD. Apart from the memory and storage improvements, these systems are dual processor (the D suffix), based on Dell 730d servers.
Having a new set of hardware products is obviously great (because increasingly, customers seem to want to purchase pre-defined hardware designs), however the real intelligence to HCI is in software. HyperCore, the HC3 operating system is upgraded to release 7.0, which introduces deduplication for the first time. Version 1 of dedupe in HC3 isn’t going to set the world alight, however 11TB of raw capacity in a system will provide around 14TB usable after data protection and dedupe is accounted for. Naturally individual customer experiences will vary, but these are typical figures. Expect to see the efficiency of dedupe increase over time as customers’ data provides more insights to tune the algorithms. Release 7.0 also introduces the ability to drive HC3 via API. The API interface has existed for some time, but wasn’t generally advertised. Now customers can automate and orchestrate HC3 directly, which opens up the possibility for scenarios like easier remote operations.
The Architect’s View
The implementation of the underlying HCI technology will drive the success of future hardware adoption. I hear interesting performance figures on HC3 and NVMe from the Scale Computing lab, for example, although these details aren’t yet public (but hopefully will be soon). As faster hardware comes to market, HCI solutions that are designed with performance in mind will be able to take advantage of the increases in speed. Scale are looking at solutions for SAP customers and clearly moving towards supporting more enterprise-class workloads. The steps to achieving this have been incremental, rather than “big bang”, which (assuming you have the financial runway) always seems like a better approach to me.
You can check out the specifications of the new platforms from the links below. Also, for background there are links to the Tech Field Day presentations and from A3 Technology Live in London.
- A3 Communications Technology Live March 2016 London: Scale Computing (Youtube, retrieved 7 August 2017)
- HC3 Systems Product Specifications (PDF, Scale Computing website, retrieved 7 August 2017)
- Scale Computing Announces Hyperconverged Appliances Designed for Storage Hungry Data Centers (Scale Computing Press Release, retrieved 7 August 2017)
- Scale Computing Presents at Storage Field Day 5 (Tech Field Day website, retreived 7 August 2017)
Comments are always welcome; please read our Comments Policy. If you have any related links of interest, please feel free to add them as a comment for consideration.
Copyright (c) 2009-2017 – Chris M Evans, first published on http://blog.architecting.it, do not reproduce without permission.