Our on-prem s3 is nightmare-ish to maintain. Should the monitoring team be looking to add VAST as the future vmstorage/backup backend? 🤔

at the finest granularity possible. Not everyone is as grounded in the reality of the ask. I love that bit from our switch vendor. Equally good - when a popular fs vendor said, "sure, you could have the best telemetry ever from our backend - but you'd get 0% fs performance." These are my people. 😜

Looks awesome! Hope you have an awesome workshop!!

Also, yay Bill! :D @nersc.bsky.social is lucky to have him. 😁

But ultimately the appeal of the field is how much can rapidly change when the underlying principles remain the same. With quantum, the underlying principles change yet again. I thrive on the lifelong learning aspect, even when new things are fast & furious. 🤩

One of my fave things about tech is that it is ever-changing, always evolving. In the trenches, I sometimes wish the release lifecycle was a bit less breakneck these days. And always approach the newness w/ the caution of cybersec, ethics, and a healthy dose of human-in-the-loop still important.

I shared with my colleague that this is arguably the best news I've ever received at the top of a ladder while pulling cable in the data center! In actuality: my personal best top-of-ladder news ever! Anyone have another fun story of receiving awesome #HPC news in-situ of the architecture? 😅

After ASI? And here I thought Agent Reese won the war!

Lots of cool advantages of off-prem / cloud too - more easy integration w/ new and emerging tech.

The fun thing about an engineering degree is that the old adage never dies, despite evolving technology - choose 2; good, cheap, fast. Warms my heart to know we're all still grounded in this foundation

Especially true when using on-prem resources - while you save on an ever-evolving "cloud cost model," the upfront $$$ amount for on-prem may looks exorbitant on outset. Same time, you've gotta price out annual licensing vs on-prem. It becomes extremely trade-off--based and nuanced as you scale.

We're here to help! But if you're a newbie breaking in, while we welcome every new person, all ideas, and are always open to improvement... my only caution is the same to any engineer as it was told to me too - "you don't know what you don't know." And in this case, historical knowledge matters. 😜

As someone with extensive training in SW engineering, systems architecture, distributed systems design, etc - I get that it seems easy until it isn't. 😜 I urge your continued engagement of the global #EEHPCWG Operational Data Analytics #ODA team. sites.google.com/lbl.gov/oda-...

Observability at this scale isn’t “just software.” It’s infrastructure... with physical interfaces, hardware faults, sensor constraints, and operational context baked in.

SW matters, but it's nowhere near the full-stack... this is hardcore complex systems architecture through-and-through.

The mission is clear: to advance science for humankind. To say that’s only achieved along a pathway of folks privileged enough to pursue a PhD is so harmful to the community at large.

When a PhD requirement eclipses you from considering others, you risk stacking a team of thought leaders that never implement anything or say the implementation is too hard until such time they delay the same commentary for the next system.

And as someone that’s spent longer in graduate school (thus having more years through the lens of impactful vs pushed-through PhDs — especially given limitations of the field in reproducibility)… necessitates a super balanced leader that can see both.

Not meant to throw shade. It’s just to illuminate inflexible thinking — i.e., an org that believes those 3 letters serve as a qualification that cannot be otherwise met via years of experience, years toward the degree, etc… there are legitimate dangers to stacking a house based on a degree.

While it’s not a simple problem to tackle, it is worth noting that it is quite refreshing to read a response that wasn’t as superficial as “APIs” 😂