Consultant with over 25 years of experience with Datacenters and Networks - connecting the blinky lights, routing the 1s and 0s.
Westerville, OH
Joined September 2008
- Tweets1.6K
- Following1.1K
- Followers413
- Likes70
After years of service with @NETIA_SA in Poland I will be removing our office service and highly recommending that clients do not entertain them as a service provider in Poland.
Contracted service with a static IP, they changed the circuit config at 22:30 on an evening with no notice at all! After jumping through hoops for remote access, we make those changes to find out that the static IP has changed. After 30+ mins on hold we are told that these updates are for the security of the network and yes, the static IP has changed. We asked for an escalation and the rep was taken back that we expected a follow up that night. We explained that if Netia can break the service at 22:30 at night they can surely fix it at this sort of time as well, to be told that the follow up will be the following business day. So this site will effectively be offline while we get updates made with our own network and those of clients. Netia is clearly not fit for purpose as a business provider!
We will recharge Netia for our time and materials to update this. Let’s see!
Going through onboarding now with @meow. Impressed so far, a personal compliance follow up.
Do they support @noble_xyz USDC yet? Anyone know?
Removing our own Circle Mint account from our current flow would streamline things a bit.
Can we run Apple hardware for @akashnet yet @gregosuri ?
I see lots of folks using the M4 variants now for AI clusters as a cost effective solution and it got me thinking…
I have to admit @TMobilePolska I am impressed. This is in the middle of the 9th largest city in Poland.
Now, considering you are trying to get me to sign a new contract with 30Mbps or 100Mbps max throughout, why would I? Also, you appear to be the only PL carrier that actually rate limits based on plan like this. What’s the point of 5G at 30Mbps?
Be safe out there… active @wirexapp phishing attempt.
Email does NOT originate from Wirex. Link is not a valid Wirex domain.
So @Bybit_Official rejected my permanent residency card for ID verification. Not only that, they claim that it is a falsified document and that no appeals are permitted.
So @sumsub perhaps you have a comment on this? You are all over us on LinkedIn and at shows to use your services, yet you incorrectly reject permanent resident cards issued in Poland as falsified documents?
@Bybit_Official
Can we run nodes on ARM yet? I believe that it was a cosmwasm issue last time I looked, but I heard that it was resolved?
Any of the super smart code folks know?
@gadikian @dylanschultzie @JoeAbbey
It has been a busy couple of weeks for @ConsensusOne!
✅ Additional Server Capacity in Europe
✅ Additional Rack in North America
Need infra that “just works”? Give us a shout!
You build, we build 💪
So, replace our n x 40G links with n x 100G? 🤔
We generally went from 2 x 10G to 2 x 25G for server to ToR, but we still have some good old QSFPs in the mix here and there, as well as a few trusty old QFX 40G switches - a few systems that previously were bandwidth heavy are 40G and some interconnect still.
Greenfield, I would go 25 and 100 all day.
I’ve spent many a winter in Edmonton…
Toques at the ready!
Feels like -46 🥶
Thankfully not this week!
What in the?!? I don’t even have words…
Ok, maybe a few now I picked myself up off the floor.
In a couple of years I can foresee Juniper buried deep in the bowels of HPE, no person of knowledge to be found, HPE partners abound, but not the ones who can help with that vertical.
😢
.@HPE and @JuniperNetworks, a leader in #AI-native networks, today announced that the companies have entered a definitive agreement under which HPE will acquire Juniper Networks. juni.pr/3TUdi3z
Given my infra and datacenter experience I love what projects like @akashnet and folks like @gregosuri are doing.
Decentralized access to GPU hardware is great - it also puts these resources within reach of an individual dev and not just a funded startup or corp.
Multi GPU workloads within a single system are easier to maintain - the GPU to GPU throughput is driven by the bus.
Curious how networks like @akashnet will offer multi GPU workload solutions that are not on the same host. These GPU to GPU workloads generate a single network flow capable of saturating a 400Gbps port.
These are the sorts of challenges we see with these workloads on the network side. Juniper has a pretty good write up on it:
community.juniper.net/blogs/…
At @CosmoverseHQ I was asked by @gadikian as to why we generally provision 2 x 25G minimum from our ToR switches and n x x00G internal Spine/Leaf. Well, these east-west requirements are getting a bit heftier.
If you administer anything with a SSH attack surface, please patch against this. Do you have to drop everything, no, but it does need to be mitigated. As this requires an active MITM attack to exploit it isn’t as straight forward to exploit over the Internet, but certainly possible. Are you vulnerable? Probably.
terrapin-attack.com
