2,165 karma · joined December 10, 2017
- MITM attacks - Ad/message injection - Malware injection - Censorship
> Furthermore, because it is self hosted there are several physical machines at different locations which serve as backups and are selected by updating the DNS records, which I believe is incompatible with a basic lets encrypt system.
Sure. But it's not that much more harder to do it. Use something like the caddy server and have it manage certificate deployment for you.
Unless there's a client that you intend to serve that absolutely cannot handle SSL/TLS, I cannot see a reason why you'd want to stick with HTTP. And no, I don't consider IE6 as a valid excuse. You can always use an older version of Firefox on Windows 2k and it'll solve this problem. I honestly don't think I can trade all the above for more compatibility with ancient machines.
> but I am solidly in the camp that if you believe your ISP is attacking you you need to use a VPN for all of your traffic, because injecting ads is the least of your worries at that point.
I don't understand this. If your resistance to switch is to make it easy for anyone with any browser to use your web page, then expecting them to use a VPN because they have a terrible ISP because you refuse to use HTTPS makes their life more difficult. Surely, they don't see this problem with other websites that use HTTPS but now need to go get a VPN service just to use yours. Also, we're assuming VPNs are absolute saints here. Not really true. But even if it were true, a VPN is eventually someone else's ISP. You're hoping nothing and absolutely nothing in the routing topology will ever look at your data or bother to inject into it.
> I think Intel / Microsoft have found a way to interconnects scalable and this comes interesting thought on value of have more cores on cpu. Especially if system was designed to be pluggable and if one of cpu's failed than it does not bring entire system down.
I think this is an interesting point. If the QPI links on the Xeon Platinum is restricted to 3, I wonder what the 32 socket mesh looks like and what implications that has on availability of the system as a whole if one CPU was to fail. Also, unless these are fully NUMA aware workloads or each socket was dedicated to a VM, the latency has to be terrible for memory access especially if the interconnect is multiple hops away. Clearly it's not going to be a fully connected mesh?
Thanks for this, btw!
EDIT: https://github.com/micro/micro/blob/master/README.md
This page is more informative on why Micro over anything else. So I think this answers my second question.
> Load Balancing - Client side load balancing built on service discovery. Once we have the addresses of any number of instances of a service we now need a way to decide which node to route to. We use random hashed load balancing to provide even distribution across the services and retry a different node if there’s a problem.
Trying to understand this a bit here. Is micro a replacement for k8s? Because I don't see any deployment capabilities here. So I'm assuming it's complementary to that. If that's the case, then I'm not sure how service discovery and load balancing would co-exist with the CNI in k8s.
What does that mean exactly? Were the people laid off given their jobs back? What does "withdrawing" a WARN notice even mean?
> // constructRoutingTree constructs routing tree according to provided routes
> // gearbox implements Gearbox interface
> // New creates a new instance of gearbox
It's obvious from the names of the functions WHAT they do. The point of the comments are to tell me WHY they are necessary. The project feels like commenting for the sake of commenting. I honestly don't know why a "routing tree" is necessary and how it's helping.
> In every case, a viable option for the customer will be to “do nothing”. They don’t have to buy your product, or any product for that matter. The best way to prove the case for your product is with your metrics. Doing nothing should be a very expensive option.
That fundamentally speaks for having a problem to solve and not a solution looking for a problem. I think this key distinction is a good sieve to filter out potential ideas especially if you're in the business of SaaS.
Their benchmarks show significant gains compared to RocksDB.
> https://github.com/spdk/rocksdb
But what I'd really like to see is a comparison against RocksDB using SPDK
> https://dqtibwqq6s6ux.cloudfront.net/download/papers/Hitachi... Based on these results, SPDK performs significantly better than the kernel requiring only 1-2 cores to saturate IOPS on an NVMe SSD (compared to the kernel requiring 16)
The article proposes using basic function call syntax. That makes no sense to me. And with .(), you now have 3 characters to type instead of 2 with []. Obviously all of this is pedantic, but then again so is the article.
> someList(0) /* instead of */ someList[0]
This is worse. How do you disambiguate that from a function call? It's now just abusing ()