Istio @Soloio_inc. More content at https://xitter.cf/t.co/tBbI542Pdw.

Joined December 2020
howardjohn retweeted
You've gotta patch your dependencies super fast because they're full of vulns (PHP, Node, eMudhra, etc), but not _too_ fast because they're full of backdoors (LiteLLM). Good luck!
23
163
24
1,841
40,097
Congrats! Blacksmith makes our CI ~5x faster, couldn't do without it
Replying to @sequoia
@sequoia is leading our Series C at a $1B valuation. We started Blacksmith in 2024 to make CI faster and less painful. In January we ran 5 million CI jobs a week. Today it's over 90 million and growing, and we’re still early.
3
91
Production AI needs more than model APIs. Agentgateway v1.3.0 adds built in cost tracking, virtual model routing, reusable providers and guardrails, plus expanded provider support, giving platform teams more control over agentic workloads from a single gateway. Read the blog by Sebastian Maniak from Solo.io: bit.ly/4y6FCBp
1
3
10
784
howardjohn retweeted
I just open sourced my "Is this slop?" simple test
114
1,018
66
18,845
475,063
Today marks exactly 6 years from my first commit of a long journey to make @IstioMesh scale. When I first started, a 100 pod mesh was a "big" mesh. Today, we can handle anything - 100 *million* pods and beyond! Service mesh scalability is a solved problem. Link 👇
2
4
1
52
5,821
Top KubeCon question: "Does ambient mesh scale?" Yes, but a more useful question is "How much effort does it take to scale"? blog.howardjohn.info/posts/s… covers this in depth.
1
14
518
howardjohn retweeted
Istio 1.24 is out today, and it's kind of a big deal. istio.io/latest/news/release…
1
12
3
33
11,289
The new compute is pretty slick... its a shame they are locked to only a few regions and any quota request is instantly denied even years after release.
Live from the App Dev & Infrastructure #GoogleCloudSummit: Our AI Hypercomputer now includes Trillium TPUs in preview, Axion-powered C4A VMs in GA, and, coming soon, A3 Ultra VMs powered by NVIDIA H200 GPUs! More on these infrastructure innovations → goo.gle/3YKibP0
2
349
Powered by Gloo Ambient mesh, I was able to push through *410k requests per second*, with mTLS encryption and HTTP monitoring. All served by a single 8-core CPU (on GCP).
1
7
25
1,891
howardjohn retweeted
"Istio was able to deliver 56% more queries at 20% lower tail latency [than Cilium]. Taking into account the resources used, Istio processed 2178 queries per core, vs Cilium’s 1815, a 20% improvement." istio.io/latest/blog/2024/am…
11
2
34
5,586
One of the hardest thing for ambient adoption has been getting people to understand how easy it is to use. Years of tricks and workarounds to make sidecars work can be thrown out since so many things "Just Work", but its hard to detach from the old ways.
Customer calls with ambient users hit different... got off a call where the issue was, I quote, "After deploying ambient cluster-wide, nothing broke and there was no latency impact at all, so something must be off". Nope... that's just how it works!
3
2
2,059