Disclaimer: I recently attended Storage Field Day 8. My flights, accommodation and other expenses were paid for by Tech Field Day. There is no requirement for me to blog about any of the content presented and I am not compensated in any way for my time at the event. Some materials presented were discussed under NDA and don’t form part of my blog posts, but could influence future discussions.
For each of the presentations I attended at SFD8, there are a few things I want to include in the post. Firstly, you can see video footage of the Pure Storage presentation here. You can also download my raw notes from the presentation here. Finally, here’s a link to the Pure Storage website that covers some of what they presented.
I wrote about Pure Storage’s new hardware platform, Pure1 and Evergreen Storage program when they were announced earlier this year. Pure’s presentation this time around wasn’t earth shattering, but they did cover off on two things that I hadn’t looked at in too much depth previously.
Hardware needs to be resilient
While we spend a lot of time in the industry talking about how software-defined storage is the future, a big part of Pure’s success has been the focus on developing a resilient, supporting hardware platform. To that end, Ryan Oler and Larry Touchette did a demo of the FlashArray//m’s hardware resiliency, and took us through some scenarios where components have failed / been removed and what happens as a result. As an aside, here’s a photo of Ryan looking extremely chuffed with the Pure logo silkscreened on a FlashArray//m.
So, when you pull a flash module (one flash module contains two SSDs), you lose about 4 – 5% of your effective capacity (there’re 20 modules in the chassis). To see what’s happening on the environment you can run “puredrive list” to get the status of any drive / module faults. When you put the flash modules back in different slots the array checks what data is still good to use. The good news here is that, in terms of recovery, the array “essentially self heals”.
When an NV-RAM module is pulled, you’ll see short spike in latency. When this happens, it writes DDR data out to internal NAND chips before it loses power.
Another useful command to run is “purehost monitor –balance”. This does a brief I/O sample on the array, which is a good way to look at host traffic balance and check that I/O is being evenly distributed across your available paths.
All of this information is pointless if you can’t make sense of it.
In much the same way that Nimble Storage are making some good use of monitoring and hardware analytics, Pure Storage are taking a similar approach with Pure1.
For Pure customers, call home data automatically generates (proactive) support tickets. Approximately 70% of the tickets in Pure are auto-generated. Pure are also making an “[a]ggressive effort to correlate and reduce noise from automatically created alerts”. They do this by processing incoming logs for known “fingerprints” to help prioritise the delivery of fixes.
Pure take the approach that “any manual process at scale is error-prone”. It’s for this reason that they want to automate the upgrade process on the arrays they sell so they can do it at scale and without human error. If you’ve ever been in a data centre at 4 in the morning on a call with a vendor’s support team after a failed array upgrade you’ll appreciate the sentiment behind this approach. This also feeds Pure’s mantra of
- Never a disruptive upgrade
- Never required a data migration
- No customer left behind
- Rapid innovation cycle
And they seem very keen on upholding this.
One of the coolest things about Pure1 Manage is that it’s SaaS. Pure can keep adding new features and capabilities without the customer needing to go around and update stuff. Pure demoed some new features that had been added since the product went GA, including adding an “Analytics” tab to the dashboard that customers could use for capacity planning.
Closing Thoughts and Further Reading
“Other” Scott Lowe had a good article on Pure Storage’s evolution that I urge you to read. I’ve been excited by what Pure has been doing for the last little while, and look forward to seeing them take things to the next level post-IPO. While people seemed to have a bit to say about the success or otherwise of Pure’s recent IPO, I think they’ve hired a lot of smart people and seem to have a solid vision and determination to execute on their current strategy. While this particular presentation didn’t set the world on fire, I think there was enough there to indicate that they’re on the right track. I’d encourage you to check them out if you’re in the market for some flash storage in your DC.