JaguarPC continuous string of networks issues: company-side or my VPS

Portal Home > Knowledgebase > Industry Announcements > Web Hosting Main Forums > Providers and Network Outages and Updates > JaguarPC continuous string of networks issues: company-side or my VPS

Posted by 1script, 02-19-2013, 12:04 PM
Hi all,

I am not having any luck with JaguarPC since late last year. It seems to be unrelenting string of network outages/latencies and overall slowness. I realize that they moved severs recently but it did nothing to improve the throughput. My next move would be to take my business elsewhere but, since I've been with them for almost 6 years, I'd like to give them a benefit of a doubt. So, I'd like to ask other JaguarPC customers here: what are your throughput rates lately?

I have been seeing 1 to 1.5MB/s through most of the last couple years until it started to go down to150kB/s, then 100kB/s often and now 60-80kB/s. All other hosting companies I have servers at are actually increasing their throughput. I've been pleasantly surprized having clocked a ServInt connection at 3Mb/s the other day, so what's the deal with JaguarPC? Bad datacenter or am I just on a bad physical server with faulty NIC or something?

I would appreciate if anyone can share their experiences with JaguarPC and network speeds, especially since the beginnnig of 2013.

Thanks!
Posted by Zachary McClung, 02-19-2013, 12:22 PM

Quote:

Originally Posted by 1script

Hi all,

I am not having any luck with JaguarPC since late last year. It seems to be unrelenting string of network outages/latencies and overall slowness. I realize that they moved severs recently but it did nothing to improve the throughput. My next move would be to take my business elsewhere but, since I've been with them for almost 6 years, I'd like to give them a benefit of a doubt. So, I'd like to ask other JaguarPC customers here: what are your throughput rates lately?

I have been seeing 1 to 1.5MB/s through most of the last couple years until it started to go down to150kB/s, then 100kB/s often and now 60-80kB/s. All other hosting companies I have servers at are actually increasing their throughput. I've been pleasantly surprized having clocked a ServInt connection at 3Mb/s the other day, so what's the deal with JaguarPC? Bad datacenter or am I just on a bad physical server with faulty NIC or something?

I would appreciate if anyone can share their experiences with JaguarPC and network speeds, especially since the beginnnig of 2013.

Thanks!

Would you please provide your ticket #? This is something we can look into and see what we can do to resolve it for you.
Posted by 1script, 02-19-2013, 10:15 PM

Quote:

Originally Posted by Zachary McClung

Would you please provide your ticket #? This is something we can look into and see what we can do to resolve it for you.

Hi Zachary, thanks for looking into this, Ticket #13708357

I just came back from a roadtrip, so not yet sure what it was all day today but I am seeing 600kB/s right this moment, so it's improved. I would still like to see the full speed even though 600 is a whole lot better than 100.

One other issue I wanted to bring your attention to: your techs are saying (repeatedly and with straight face, apparently) that if they clock the connection at 148MB/s , then the issue is fixed. Can someone please exlpain to them that this can only be possible over a private network, and not a shabby one at that! If they are this good with understanding how their own network is laid out, how are they expected to be able to fix issues with it? The assurances that they are connected just as their customers are just ridiculous - I don't even have 20Mb/s (NOT 20MB/s - 8 times less than that!) coming in, how can they possibly be connected the same way? They must be using some sort of a cashing proxy which "shields" them from the effects of networks slowness by serving a local copy of the files.

So, please have real techs to take a look at this,

Thanks!
Posted by Zachary McClung, 03-04-2013, 05:40 PM

Quote:

Originally Posted by 1script

Hi Zachary, thanks for looking into this, Ticket #13708357

I just came back from a roadtrip, so not yet sure what it was all day today but I am seeing 600kB/s right this moment, so it's improved. I would still like to see the full speed even though 600 is a whole lot better than 100.

One other issue I wanted to bring your attention to: your techs are saying (repeatedly and with straight face, apparently) that if they clock the connection at 148MB/s , then the issue is fixed. Can someone please exlpain to them that this can only be possible over a private network, and not a shabby one at that! If they are this good with understanding how their own network is laid out, how are they expected to be able to fix issues with it? The assurances that they are connected just as their customers are just ridiculous - I don't even have 20Mb/s (NOT 20MB/s - 8 times less than that!) coming in, how can they possibly be connected the same way? They must be using some sort of a cashing proxy which "shields" them from the effects of networks slowness by serving a local copy of the files.

So, please have real techs to take a look at this,

Thanks!

Sorry I missed your reply here. We made some changes network wise around this time and are making some additional changes in the coming weeks. Are you continuing to see the better speeds?
Posted by 1script, 03-04-2013, 06:32 PM

Quote:

Originally Posted by Zachary McClung

Sorry I missed your reply here. We made some changes network wise around this time and are making some additional changes in the coming weeks. Are you continuing to see the better speeds?

Hi Zachary,

It was kinda coming and going all this time - I would clock it at 200kb/s, then 800, then 100 again. I've been adding to that support ticket I referred to, including MTR traces from me to the VPS.

Last 24 hours I think it's been pretty stable at 1Mb/s except of course last evening the VPS was down for 2 hours - noone could tell me why the VPS was not brought up when the hardware node needed to reboot?

Anyhow, the downtime is a separate issue and 1Mb/s is a-OK speed, I can live with that if it holds. But, since I can download from my other VPSes with some of your competitors at 3+Mb/s, I have to ask: what would you consider "normal" throughput to end users on the US East Coast?
Posted by 1script, 03-10-2013, 10:04 PM
Zachary, since the issue is still not permanently fixed (it's been about 24 hrs of good 1.5MB/s speeds and just slid back to 200-300kB/s today), I would like you to respond publicly and find some holes in my logic talking to your support stuff. They still keep saying that evertything is allright because they can get the test file at 40MB/s+ which is insanely and improbably fast over the actual Internet. What do you personally make of my exchange with your techs?

Quote:

I am running the download test now, it's still downloading and coming at about 200kB/s, I'll post update when it's done (might take a while).

HOWEVER:

Sreejith, Samuel and everyone else involved: please, please: download speeds of 41.4 MB/s or 26.1 MB/s or even anything really above 5-6MB/s are unattainable over the actual Internet which normal people like most of the visitors to the sites of your customers are having. When you send me your test results with speeds like that, what you are essentially saying is that you are connected over a dedicated (most likely MPLS) channel which most likely is private and therefore almost guaranteed to have separate sets of settings (routing, firewal rules etc.) from what I and the rest of my REAL users are seeing over an actual Internet.

Only in the last year or so Verizon FIOS has started to introduce 50Mb/s (bits) fiber channels to actual REAL subscribers and that's actually only 6.25MB/s (bytes) not considering any bottlenecking, multiple router delays beyond your network etc.

So, please stop referring to your own download speeds as valid speed tests.

OK, my test download is done (303kB/s):
<private details snipped>

Do they have the right idea about how they are connected to your network?
Posted by Flapadar, 03-13-2013, 07:22 PM
If your server can connect to another server at 300-500mbps, the problem is verizon or whoever you use. Or the latency between you and the server: which is mostly just affected by distance. TCP overhead causes slowdowns as latency increases.
Posted by 1script, 03-13-2013, 09:18 PM

Quote:

Originally Posted by Flapadar

If your server can connect to another server at 300-500mbps, the problem is verizon or whoever you use.

Server to server connection is pretty much irrelevant - I'm interested in the speed that normal people can connect to it, and that's always been the key. I don't care how awesome are JaguarPC's own staff connections to their servers I just want 6 million of my potential customers (Verizon alone) to have no lag.

Regardless, on the same Verizon connection I download from my other servers hosted by Jaguar's competitors at the other side of the US, at 10 times the speed.

I'm working with Jag on the resolution, BTW, hopefully it will come one day ...
Posted by Zachary McClung, 03-13-2013, 10:20 PM

Quote:

Originally Posted by 1script

Server to server connection is pretty much irrelevant - I'm interested in the speed that normal people can connect to it, and that's always been the key. I don't care how awesome are JaguarPC's own staff connections to their servers I just want 6 million of my potential customers (Verizon alone) to have no lag.

Regardless, on the same Verizon connection I download from my other servers hosted by Jaguar's competitors at the other side of the US, at 10 times the speed.

I'm working with Jag on the resolution, BTW, hopefully it will come one day ...

Our network admins continue to work on the issue; however, I do not know how much we are going to be able to do on it. Based on the results you provided to us, the issue is outside of our network and it is an issue between Verizon and nlayer. The speed test the admins had provided were between the two Atlanta facilities. It was to confirm that the issue was not within our network. As the MTR also shows it is outside of our network. They are still looking to see if they can improve speed through working with nlayer. They will update you once you have more information.
Posted by 1script, 03-14-2013, 12:01 AM

Quote:

Originally Posted by Zachary McClung

Our network admins continue to work on the issue; however, I do not know how much we are going to be able to do on it. Based on the results you provided to us, the issue is outside of our network and it is an issue between Verizon and nlayer. The speed test the admins had provided were between the two Atlanta facilities. It was to confirm that the issue was not within our network. As the MTR also shows it is outside of our network. They are still looking to see if they can improve speed through working with nlayer. They will update you once you have more information.

Zachary, thanks for the update. I understand that the issue may not necessarily be in the network that literally belongs to you but unfortunately it does not make it "not your issue". I cannot complain to GTT about issues with nlayer - I'm not their customer. Perhaps you are not their direct customer either but you've got to become a champion of my complaint and push the issue up the chain. I don't know how many layers of command/ownership we have to penetrate but there is a device out there that appears to be failing or becoming overloaded and the actual owner needs to look at it.

The slowest part of that MTR is actually in the atlas.cogentco.com network, far away from Verizon and only two hops from you. You have my MTRs, the IP's are right there (well, OK, hosnames - I can get you IPs as well if needed, all in 154.54.x.x range) Find who owns the slowest device and have them take a look at it. Or find the guy who pays money to the guys that pay money to the guys that own the device and raise hell with them. That's the only way it works with ISPs, I should not be explaining this to you, you know it better.

I've worked for a CLEC years ago, that's how we dealt with call quality issues. Find the first guys you owe money for the service to and withhold the payment. That makes the ticket go up the chain real fast. And, by the way, you've already got my payment for next month two days ago even though my sites are still running at 200kB/s, so you can see I put a lot of trust in you, based on the previous 6 years I've been with JaguarPC.

Keep me posted please here or in your ticket system.
Posted by 1script, 06-01-2013, 08:23 PM
The issue had cleared two months ago but three days ago had creeped up again. It's been progressively worse and I'm down to just a trickle of about 80kB/s - absolutely inadequate.

I have submitted a ticket, referenced the last ticket that was resolved, and marked my ticket URGENT. Noone has looked at it in 24hrs.

This just has to be a JaguarPC - wide issue because of the MTR traces I'm getting, I am amazed you haven't excalated it up to your ISP provider yet.

Zachary, what does it take to have someone at JaguarPC to look at URGENT tickets in less than 24hrs?
Posted by kpmedia, 06-02-2013, 12:58 AM

Quote:

Originally Posted by 1script

This just has to be a JaguarPC - wide issue

Nope - no issues here.

.
.
Posted by MattF, 06-02-2013, 01:33 AM

Quote:

Originally Posted by 1script

Zachary, thanks for the update. I understand that the issue may not necessarily be in the network that literally belongs to you but unfortunately it does not make it "not your issue". I cannot complain to GTT about issues with nlayer - I'm not their customer. Perhaps you are not their direct customer either but you've got to become a champion of my complaint and push the issue up the chain. I don't know how many layers of command/ownership we have to penetrate but there is a device out there that appears to be failing or becoming overloaded and the actual owner needs to look at it.

The slowest part of that MTR is actually in the atlas.cogentco.com network, far away from Verizon and only two hops from you. You have my MTRs, the IP's are right there (well, OK, hosnames - I can get you IPs as well if needed, all in 154.54.x.x range) Find who owns the slowest device and have them take a look at it. Or find the guy who pays money to the guys that pay money to the guys that own the device and raise hell with them. That's the only way it works with ISPs, I should not be explaining this to you, you know it better.

I've worked for a CLEC years ago, that's how we dealt with call quality issues. Find the first guys you owe money for the service to and withhold the payment. That makes the ticket go up the chain real fast. And, by the way, you've already got my payment for next month two days ago even though my sites are still running at 200kB/s, so you can see I put a lot of trust in you, based on the previous 6 years I've been with JaguarPC.

Keep me posted please here or in your ticket system.

So you managed to determine the bottleneck effecting you in a network path based on an ICMP traceroute?

I am sure the business lesson is appreciated though.
Posted by 1script, 06-02-2013, 12:06 PM

Quote:

Originally Posted by MattF

So you managed to determine the bottleneck effecting you in a network path based on an ICMP traceroute?

Only way I could do it as a client with no admin rights on any single device on this patch except for the beginning and the end. If you have any other suggestions, I'm all ears. Jaguar techs have requested the MTR traces themselves, so I have to assume they too use it as a starting point.

Come to think of it, I would have no qualms about JaguarPC simply discarding whatever I told them and doing the fix as they see fit. But that's not happening, hence this post.
Posted by 1script, 06-02-2013, 12:21 PM

Quote:

Originally Posted by kpmedia

Nope - no issues here.

What ISP you're coming from if you don't mind my asking? I have absolutely no idea about the routing and connections at the Atlanta datacenter this VPS is at and so I think it's possible that bandwith to different destinations is affected in a different way. So, yes, that statement of mine should be reworded as "all JaguarPC customers hosted in Atlanta that care about traffic from Verizon FIOS network". That's about 6 million potential visitors from US. I think it's a large enough cause for concern.

Ontop of the technical issues, my support ticket marked "Emergency" is still unassigned 14 hours since it's been posted. Some emergency... Hence my activity in this forum. I would much rather deal with Jaguar techs directly if they responded.
Posted by Jeffreyw, 06-05-2013, 11:22 AM
I'm glad am done with Jaguar PC, so they still have this recurring problem.