4 ms·
> ...in 2018, AWS outperformed GCP by 40%, which we attributed to AWS’s Nitro System present in c5 and m5 series... In 2020, we see a return to similar overall
by ignoramous 7y ago
> ...in 2018, AWS outperformed GCP by 40%, which we attributed to AWS’s Nitro System present in c5 and m5 series... In 2020, we see a return to similar overall performance in each cloud.
How did GCP and Azure catch-up to AWS Nitro in one year which IIRC is a coming together of minimalistic micro-vms, hardware-accelerated network and IO cards, hardware offload for encryption and other maintenance tasks... a work that was 5 years in the making [0]?
[0] https://www.youtube-nocookie.com/embed/rUY-00yFlE4?t=1m45s https://www.youtube-nocookie.com/embed/rUY-00yFlE4?t=1m45s
- zeroxfe 7y agoIt's more likely that all of them started working on it at about the same time and AWS came out first.
- cdoxsey 7y agoMeltdown and Spectre did a real number on the c5 and m5 instance types. Amazon's original fix reduced performance substantially, which lot's of people noticed, but no one knew why till the vulnerabilities were announced. Meanwhile Google built a clever workaround which minimized the pain. It was really a mic-drop moment from Google and I wish someone would dig in deep and tell the whole story. At this point I'm sure AWS has merged Google's patches, but it showed how much Google was investing in their GCP offerings.
- DDerTyp 7y agoInteresting! Do you know of any resource where I can read more about this?
- campers 7y agoThere was a few interesting bits in a post the other days about their new compute VM class, #4 in particular https://cloud.google.com/blog/products/compute/understanding-dynamic-resource-management-in-e2-vms https://cloud.google.com/blog/products/compute/understanding... Its pretty light on details but was news to me.
- 9nGQluzmnq3M 7y agoThis is not related to Spectre/Meltdown, which was over a year ago.
- 9nGQluzmnq3M 7y agoThe tech is called Retpoline, and this blog has some pointers: https://www.blog.google/topics/google-cloud/protecting-our-google-cloud-customers-new-vulnerabilities-without-impacting-performance/ https://www.blog.google/topics/google-cloud/protecting-our-g...
- libso 7y agoIt is less to do with Google investing in GCP offerings than their investment in general in Operating Systems and their sheer quality of engineers.
- ladzoppelin 7y agoWhich I am sure will also work at Amazon or have worked at Amazon.
- deadmutex 7y agoMaybe
- 3fe9a03ccd14ca5 7y agoUnlikely.
- daxfohl 7y agoThe writeup states that their observation is that compute perf is the biggest factor in the TCP-C benchmark. Nitro AFAIU is mainly about networking. I think all the clouds use the same basic physical CPU sets, so I'm more surprised about the perf being much different in previous years, than I am about the perf being similar this year. Maybe the hypervisor layer was much more efficient at AWS or something?
- panarky 7y ago> GCP’s network looks much better than either AWS or Azure’s networks. > Not only do their top performing machines beat each network’s top performing machines, but, so to do their bottom performing machines. Even their least performant machine (n1-highcpu-16 in figure 10) is consistent with AWS’ maximum network throughput as seen in our tests. > This is especially impressive because last year, AWS outperformed GCP in our network tests. It is a credit to GCP that they have improved their network performance and we are left wondering exactly how they accomplished this improvement.