Re:Datapath.io: Provider-Independent Elastic IP Addresses
by amenonsen on Thursday January 29, 2015 @01:13AM
(#8964613)
When I first saw the page (opened along with a bunch of other tabs from HN links), with its "Timeout error!" balloon and "Service disruption" headline, I thought it was a fancy error page and hit reload. Oops.
Re:Datapath.io: Provider-Independent Elastic IP Addresses
by hhw on Thursday January 29, 2015 @12:47AM
(#8964546)
Doesn't every CDN already accomplish the same functionality with multiple origins and failover detection, while accelerating static content by caching it at the edge? Given how cost competitive CDNs already are, I don't see how they can charge a low enough cost in comparison to be compelling, while still incurring bandwidth costs and at least some network infrastructure which make it almost as expensive to run, minus some caching nodes. And if for whatever reason, there does happen to be some use case for anycasting the IP that's advantageous, any CDN could add this feature overnight while having a much more established footprint and infrastructure in place.
Re:Datapath.io: Provider-Independent Elastic IP Addresses
by eof on Wednesday January 28, 2015 @08:23PM
(#8963780)
> If your application goes down, all you want is to get it back up. What if your cloud provider is down? Use DATAPATH.IO to route your users to a healthy location while keeping your IP addresses.
Re:Datapath.io: Provider-Independent Elastic IP Addresses
by mmastrac on Wednesday January 28, 2015 @08:32PM
(#8963802)
I think the idea is that datapath only functions when you go down. ie: if only you go down, datapath will fix it via BGP magic. If datapath is down and you are up, stuff continues to work.
Re:Datapath.io: Provider-Independent Elastic IP Addresses
by bdonlan on Wednesday January 28, 2015 @11:32PM
(#8964358)
In general, most ISPs will drop any routes more specific than a /24. Presumably datapath has negotiated with some specific providers to allow /32s; it's up to that ISP to scale their routers to handle it; other ISPs will only see the /24 aggregate.
Re:Datapath.io: Provider-Independent Elastic IP Addresses
by rev_bird on Wednesday January 28, 2015 @11:56PM
(#8964425)
> What if your cloud provider is down? Use DATAPATH.IO to route your users to a healthy location while keeping your IP addresses.
Call me crazy, but I'm not sure I really care about keeping my IPs. Route 53 seems as close to bulletproof as we're going to find with DNS, and it's got cloud-independent failover routing built in. Point the primary record at your nice little EC2 server, and have the secondary record pointing as Azure or Google Cloud or your mom's laptop, whatever, and it'll do what Datapath is doing without futzing with your routing. ...right?
Re:Datapath.io: Provider-Independent Elastic IP Addresses
by sspies on Thursday January 29, 2015 @02:54AM
(#8964867)
Id adds one hop (our appliance) to the path. It tries to increase performance of your path by taking non-standard BGP metrics into account: congestion and latency
Re:Datapath.io: Provider-Independent Elastic IP Addresses
by sspies on Thursday January 29, 2015 @03:18AM
(#8964932)
Traffic passes the appliance at the edge of the hosting provider. It takes congestion at all links and the appliance itself into account and re-routes accordingly.
Depending on the characteristics of the location, at least three very unsimilar transit connections are used.
Re:Datapath.io: Provider-Independent Elastic IP Addresses
by sspies on Thursday January 29, 2015 @03:41AM
(#8964997)
Transit providers sell access to the whole internet to us. We are in negotitations with them, but we will be more specific, once we have the commitments. Technically, we choose at least three different providers, so our appliance has a good basis of decision-making.
For the connection between your VPC and our appliance, we use the regular AWS DC API.
The "vague bullshit in between cute animations" approach really doesn't work at all for this type of product.
EDIT another: https://twitter.com/datapath_io/status/559741202342625280/ph...
But... what happens if datapath.io goes down?
https://www.youtube.com/watch?x-yt-ts=1422411861&feature=pla...
But, would this not break the internet if everyone used it? routing individual IPs (rather than large blocks) sounds like it would end in chaos?
Call me crazy, but I'm not sure I really care about keeping my IPs. Route 53 seems as close to bulletproof as we're going to find with DNS, and it's got cloud-independent failover routing built in. Point the primary record at your nice little EC2 server, and have the secondary record pointing as Azure or Google Cloud or your mom's laptop, whatever, and it'll do what Datapath is doing without futzing with your routing. ...right?
How is the TTL not reliable?
In my personal experience DNS cut-overs have never been a problem. Is there any major OS or ISP doing DNS wrong?
Where is it physically located, what about congestion of the appliance itself, what are your peerings?
What does transit mean in this context; like between my VPC and your appliance? Can you give an example for e.g. an AWS region?
How is the appliance implemented, is it physical hardware, EC2 instances? Is it redundant, how do you scale it in terms of bandwidth?