It is easy to kick someone who is down. Bad news have stronger ripple effects than the good ones. Intel® is going through a rough patch, and perhaps the worst one so far. They delayed their 7nm manufacturing process, one which could have given Intel® the breathing room in the CPU war with rival AMD. And this delay has been pushed back to 2021, possibly 2022.
Intel Apple Collaboration and Partnership started in 2005
Their association with Apple® is coming to an end after 15 years, and more security flaws surfaced after the Spectre and Meltdown debacle. Extremetech probably said it best (or worst) last month:
If we look deeper (and I am sure you have), all these negative news were related to their processors. Intel® is much, much more than that.
Their Optane™ storage prowess
I have years of association with the folks at Intel® here in Malaysia dating back 20 years. And I hardly see Intel® beating it own drums when it comes to storage technologies but they are beginning to. The Optane™ revolution in storage, has been a game changer. Optane™ enables the implementation of persistent memory or storage class memory, a performance tier that sits between DRAM and the SSD. The speed and more notable the latency of Optane™ are several times faster than the Enterprise SSDs.
Intel pyramid of tiers of storage medium
If you want to know more about Optane™’s latency and speed, here is a very geeky article from Intel®:
The list of storage vendors who have embedded Intel® Optane™ into their gears is long. Vast Data, StorOne™, NetApp® MAX Data, Pure Storage® DirectMemory Modules, HPE 3PAR and Nimble Storage, Dell Technologies PowerMax, PowerScale, PowerScale and many more, cement Intel® storage prowess with Optane™.
3D Xpoint, the Phase Change Memory technology behind Optane™ was from the joint venture between Intel® and Micron®. That partnership was dissolved in 2019, but it has not diminished the momentum of next generation Optane™. Alder Stream and Barlow Pass are going to be Gen-2 SSD and Persistent Memory DC DIMM respectively. A screenshot of the Optane™ roadmap appeared in Blocks & Files last week.
Intel next generation Optane roadmap
DAOS (Distributed Asynchronous Object Storage)
Intel®’s foray into SC (supercomputing) and HPC (high performance computing) when they acquired Whamcloud in 2012 looked like a fish-out-of-the-water move. The Lustre clustered parallel file system always had a difficult and convoluted history of ownerships and developments, and for me at the year of the Whamcloud heist felt weird. In a series of events I forgot, I recalled having a drink with 3 Intel® folks who requested for the meeting. Avinash Rajah, Chris Sassone and Malcolm Cowe may have found me through my SNIA® connections, and soon after, I was picking up the Lustre stuff – architecture, installation, and deployment. Intel® was looking for partners to resell the Lustre L3 support and that was their business model until they finally sold off Lustre to DDN® 2 years ago. I enjoyed Lustre, having some design and consulting work, and one implementation job as well.
The Lustre pass off brought in the Intel® DAOS technology, an exascale level high performance storage for supercomputing. Already DAOS is returning the benefits of Intel®’s strategy, having topped recent IO500 benchmark and doubled its number over its predecessor, WekaIO.
Intel® recently presented at Storage Field Day 20, and this video (around the 11.30 mark), talked about the DAOS technology. It impressed SFD20 delegates a lot, because I see several of them posting their positive views on social media.
Intel Optane Persistent Memory Usages from Gestalt IT on Vimeo.
The million dollar question is how will Intel® profit from DAOS technology? Going down the L2/L3 software support route has the uncertain repercussions of the Lustre past, and the freemium model of open source/enterprise features isn’t exactly Intel®’s strength in engaging the enterprise and hyperscale market as well.
There was a DAOS presentation at SNIA SDC 2015.
The Ruler
This format, the EDSFF (Enterprise & Datacenter SSD Form Factor) is another game changer. The capacity density, with the benefits of rack space, power and cooling savings, has completely changed the way we look at storage medium form factors.
The Intel “Ruler” EDSFF form factor
The “Ruler” form factor was introduced in 2017, and it has garnered big names support ever since. Intel® (obviously), Samsung, Microsoft® Project Denali, Supermicro®, Lenovo®, Kioxia (formerly Toshiba Memory) and many more are jumping into the bandwagon of the next generation of storage medium form factor.
Giving credit where it is due but …
I am not paid to write this. I have no obligations to appease anyone and I am writing to give credit where credit is due. In my engagements and observations over the years, there are a lot of very smart people at Intel®. They have incredible technologies related to storage, but has yet to create a progressive ecosystem like what they have done with their processors for decades. On the hardware side, they might be able to lean on their experience and history to build the connection with the OEMs, ODMs, supply chain and so on.
But it is the software side where I have not seen Intel®’s best to be the best. I hope DAOS will buck that trend and lead Intel® to be a great software company. [Note: Clear Linux* is another one on my watch list].
And it also boils down to execution of the business strategies and how Intel® attacks the storage market. It must not continue to fumble and tumble, and it is time change that perception because Intel® can be dominant beyond their processors too.
[ Update Aug 16, 2020: After I finished this blog, this article came out. It was apt to what I have written about Intel® and its storage innovation ]
Pingback: Intel Is Still a Formidable Force - Tech Field Day