Most active commenters
  • boulos(5)

←back to thread

343 points cvallejo | 11 comments | | HN request time: 1.188s | source | bottom
1. boulos ◴[] No.22357462[source]
Disclosure: I work on Google Cloud.

cvallejo is the PM, so ask her anything!

replies(3): >>22357543 #>>22359291 #>>22360897 #
2. JoshTriplett ◴[] No.22357543[source]
Does the new n2d machine type support nested virtualization?

(Asking because Azure supports nested virtualization but only on some machine types. AWS doesn't support nested virtualization at all. Google Cloud seems to support nested virtualization on other machine types.)

Also, why 224 rather than 256?

replies(2): >>22357620 #>>22357757 #
3. boulos ◴[] No.22357620[source]
We do not yet support AMD's nested implementation (we do on Intel). But cvallejo is also the PM for Nested :).

As for 224, we've always reserved threads on each host for I/O and so on. Figure 2 from the Snap paper [1] is probably the best public reference. We also don't make it clear (on purpose) what size the underlying host processors are, though you can clearly guesstimate pretty easily.

[1] https://research.google/pubs/pub48630/

replies(2): >>22357723 #>>22357725 #
4. QUFB ◴[] No.22357723{3}[source]
Speaking of nested virtualization, are there any plans to support this on E2, or will we have to use other instance types?
replies(1): >>22359201 #
5. JoshTriplett ◴[] No.22357725{3}[source]
> We do not yet support AMD's nested implementation (we do on Intel).

Any particular reason for that limitation, or just "not implemented yet"? (Not asking for product roadmaps, just wondering if there's a specific technical issue that makes it more difficult to support.)

> As for 224, we've always reserved threads on each host for I/O and so on. Figure 2 from the Snap paper [1] is probably the best public reference.

That's helpful, thank you.

replies(1): >>22360134 #
6. ◴[] No.22357757[source]
7. boulos ◴[] No.22359201{4}[source]
The blocker here would be the need for supporting nested on AMD parts (as E2 can choose between Intel and AMD parts).
8. outworlder ◴[] No.22359291[source]
Are those instances available to all customers yet? We don't see those in our console.
replies(1): >>22360743 #
9. boulos ◴[] No.22360134{4}[source]
Not implemented yet. In the stack rank of “stuff needed to update our hypervisor for AMD again” it wasn’t at the top :).

Note the again as well: GCE originally had it such that N in N1 meant iNtel, and A1 was for AMD (as Joe said publicly here: https://twitter.com/jbeda/status/1159891645531213824). By the time I joined though, we didn’t see the point of the A1 parts, since the Sandybridge’s smoked them.

10. boulos ◴[] No.22360743[source]
See my other comment about the rollout proceeding now. You should see them sometime this week, in theory.
11. ksec ◴[] No.22360897[source]
I am pretty sure it is not Your / Google Cloud's fault given AWS and Azure are in similar situation here, but;

ROME was announced in Later 2018, Released on Aug / September 2019. Why did it take another 6 months to roll out an new instance? When I assume you must have had samples since early 2019.