The Israeli tech news site, Globes, reported today that Amazon is acquiring the NVMe flash storage start-up, E8. The report mentions the deal is estimated to be between $50 and $60 million. However, the article also notes that other sources estimate the deal might have been for less.
So, why is Amazon buying E8?
At this time, I have not seen an official statement from Amazon. I speculate, however, that this is a technology, rather than a business, acquisition, something that offers high performance NVMe storage that fits Amazon’s hyperscale architecture. E8’s technology may be delivered as an ultra-fast storage tier. Or Amazon may use the technology in future versions of Outposts. Either presents a fascinating opportunity for Amazon and its customers.
I do not, however, have to speculate on E8’s technology. We at ESG evaluated E8’s technology a couple of years ago.
Here are some of the highlights from our report.
- E8’s design leverages standard, readily available hardware. The solution we validated was a 2U enclosure supplied by an Intel whitebox ODM.
- The E8 solution that ESG validated leveraged RDMA over Converged Ethernet (RoCE) for high performance, low latency storage network performance.
- The E8 client software on the host server handles multi-pathing, including path failover, and can accommodate dual Ethernet switches for network redundancy. This technology means Amazon doesn’t have to wait for NVMe multi-pathing support to solidify.
- E8 delivers incredibly efficient storage performance. In our testing, E8’s software, along with the RoCE protocol, added only about 10us of additional latency in a host request on an NVMe drive. The 2U E8-D24 storage array was able to deliver over 10 million IOPS for a 100% random read workload and over 43 GB/sec of throughput for large block read operations.
What does this mean going forward?
Amazon got a solid, high performance storage technology that, if the article’s estimate is accurate, was still relatively cheap. I assume that given E8’s start-up nature, Amazon does not have a ton of established on-premises customers to support. When I compare this to Google’s move for Elastifile that I wrote about few weeks prior, there are both similarities and differences.
In both cases, a public cloud service provider acquired an on-premises focused storage startup to augment their storage offerings. This alone could change the storage market; there is a new crop of tech buyers out there. It will be interesting to see if the venture capital community gets more bullish of storage startups.
From the outside, Elastifile makes slightly more sense. Both are excellent technologies, but high-performance scale-out file systems are incredibly complex to build and could be considered a weakness among current cloud storage portfolios. While E8’s performance efficiency is stellar, Amazon might have been able to build something similar, or good enough, by leveraging newer hardware with some adjustments to their existing technology.
Does E8 represent a strategy shift at Amazon from build to buy for storage technologies? Or is the rising interest in high performance workloads, such as machine learning, driving Amazon to seek out performance advantages wherever it can? Time will offer more insight. In the meantime, E8 is solid technology and I expect Amazon to put it to good use.
