4 ms·

> ...all on the strength of network admins who didn't want to learn anything new to move to the cloud. Anyone who thinks this way is super cynical, including C

> ...all on the strength of network admins who didn't want to learn anything new to move to the cloud. Anyone who thinks this way is super cynical, including Cisco salespeople. Isn't "keep the same interface when you move to the cloud" a desirable thing? This is like saying "People who use RHEL AMIs on AWS are lazy because they don't want to learn anything new, they should just migrate en masse to Lambdas or some other cool new tech". It's ridiculous that we bemoan the state of moving-every-night Javascript framework on HN and also claim that network admins "don't want to learn anything new" just because they pay Cisco for working AMIs. I think you're just jealous that network admins don't have to deal with the messes of Next.js and Kubernetes ;) I wish I could get the stability of the Cisco CLI in my day-to-day work on cloud infrastructure.

wmf 3y ago
Keep the same interface at what cost? Public cloud is already expensive and Cisco-on-AWS or VMware-on-AWS are really expensive. In many cases customers are buying the same functionality twice which feels exploitative.
- quadrifoliate 3y ago
  I don't think anyone is buying Cisco-on-AWS at anywhere near the same volume as Cisco gear in its heyday. It's likely for applications where traffic throughput is low or unimportant — e.g. training labs, etc. These are likely applications where it would have been cost prohibitive to buy a "real" Cisco router. I'd be pretty surprised if people were running an actual large production network on Cisco-AMIs-on-AWS.
- numpad0 3y ago
  Is people paying amounts of money to a company is the problematic part? I mean, I get it, but that is not enough literal justification to single out Cisco Systems from myriad other large corporations.
jiggawatts 3y ago
> Isn't "keep the same interface when you move to the cloud" a desirable thing? NO. I wish I could somehow bold that and make it 100 pt size. The cloud is the interface. It’s a distributed multi-tenant operating system with accounts, ACLs, event logs, scheduled tasks, processes (lambda), and so forth. Most critically: it has a fully virtualised software defined network already! Layering anything on top is absolute madness. It’s complex enough as it is. The built-in networking is integrated into the monitoring, has the same RBAC model as everything else, and can be templated out using a uniform API and declarative deployment syntax. Something like Cisco will sit on top of that like a steaming hot turd on a manicured lawn. Last but definitely not least: the cloud-native network is hardware accelerated into FPGAs (Azure) or custom silicon (AWS). There is absolutely no way for a vendor like Cisco to even begin to approach the scalability or the latency of the native network in software. A customer of mine has made the catastrophic mistake of adopting Aviatrix instead of educating their network team (or simply making them redundant.) Aviatrix introduced a latency overhead of 3-5ms within the same zone compared to the native latency of 80 microseconds! That’s a brutal amount of overhead for everyone else to pay for “our network staff act like dinosaurs before the impact and I’m too scared to tell them their careers are over”.

wmf 3y ago

Keep the same interface at what cost? Public cloud is already expensive and Cisco-on-AWS or VMware-on-AWS are really expensive. In many cases customers are buying the same functionality twice which feels exploitative.

quadrifoliate 3y ago
I don't think anyone is buying Cisco-on-AWS at anywhere near the same volume as Cisco gear in its heyday. It's likely for applications where traffic throughput is low or unimportant — e.g. training labs, etc. These are likely applications where it would have been cost prohibitive to buy a "real" Cisco router. I'd be pretty surprised if people were running an actual large production network on Cisco-AMIs-on-AWS.
numpad0 3y ago
Is people paying amounts of money to a company is the problematic part? I mean, I get it, but that is not enough literal justification to single out Cisco Systems from myriad other large corporations.

jiggawatts 3y ago

> Isn't "keep the same interface when you move to the cloud" a desirable thing? NO. I wish I could somehow bold that and make it 100 pt size. The cloud is the interface. It’s a distributed multi-tenant operating system with accounts, ACLs, event logs, scheduled tasks, processes (lambda), and so forth. Most critically: it has a fully virtualised software defined network already! Layering anything on top is absolute madness. It’s complex enough as it is. The built-in networking is integrated into the monitoring, has the same RBAC model as everything else, and can be templated out using a uniform API and declarative deployment syntax. Something like Cisco will sit on top of that like a steaming hot turd on a manicured lawn. Last but definitely not least: the cloud-native network is hardware accelerated into FPGAs (Azure) or custom silicon (AWS). There is absolutely no way for a vendor like Cisco to even begin to approach the scalability or the latency of the native network in software. A customer of mine has made the catastrophic mistake of adopting Aviatrix instead of educating their network team (or simply making them redundant.) Aviatrix introduced a latency overhead of 3-5ms within the same zone compared to the native latency of 80 microseconds! That’s a brutal amount of overhead for everyone else to pay for “our network staff act like dinosaurs before the impact and I’m too scared to tell them their careers are over”.