Hosting reviews / questions

Just as I assumed... Those tests clearly aren't fair - the OVH box has been sitting at 100% for at least twice as long as the Zare machine alone.

You should also read what I said. I clearly said

but gl;hf getting a Zare 7700k to clock faster than a OVH 4790k with areal-world load

We can't even see what load you're testing on the Zare 7700k machine, the screenshot just happens to cuts that off... Memory generates heat, disk generates heat, etc.

I do accept - Zare could have changed things since I was last with them[snip]

I've just spoken to Ben, he should be running a 15 minute test on each box and post back here soon.

Edited Ben went AFK, but when I got back I did some testing and got these results. These are honestly great results to see from Zare, and it's given me more confidence in their products. I'll correct my initial post - I do admit I was wrong in stating this is an issue you can expect with Zare. I was stating this from past experience, and I should have made that clear.

Zares lowest temps are just about OVH's peak temps, giving the Zare machines ~25° leg-room before you would see thermal throttling in these test conditions. Still, OVH knock the ball out the park - so I have to say I'll be sticking with them.

So for the longest of time I've always thought the Canadian Servers where hosted more North in Canada which was because of iffy ping in the west (but doesn't really go over 80 anymore for some reason, checked this a good 2 weeks ago when I had a server w/ crident for a while) but now since they're labeling it it seems the Canadian servers actually favor the US East a lot, which is a good thing because most of the gmod population is in the East.

Although US West isn't being favored as much, the pings are still rather decent, you really shouldn't notice a difference in gmod anyways, especially if you run a lower than 33 tickrate server (even 33 tickrate would probably make it unnoticeable).

So I think it's safe to say that it's okay to switch off your dedicated US East hosts and switch to a host using OVH, cause even though their support has been absolute dog shit, shared hosting gives better support than them, especially Crident.

Crident actually made a map for their services which is which any other service provider using OVH would have too: Hopefully that can give you an idea of what I'm talking about with the US East not having that much of a problem

Now onto a review I forgot to post in the OP, but since I'm already posting why not just have it here:

#GMCHosting

I may be a bit bias when it comes to their web hosting because I've tested it out, and it was absolutely terrible, we're talking slow as fuck, according to etrug mybb crashes their web servers because of the php inside it? But all around the web hosting is a disaster and I'm really glad I stuck to my VPS instead. Onto the game hosting though: I used them for a good month or so and honestly received near terrible pings in the West Coast (even my ex in the east coast of PV got over 80 ping somehow in the "US East" location) it was just a disaster, no one would join the server cause obviously 200 ping in the west coast and triple digits in the East won't get anyone to join because people care about ping, but it was an issue mainly cause it was noticeable ping issues which is tbh pretty hard to achieve in gmod, even with 150+ ping on a UK server I staffed on, I was nearly never affected by it. It may have had something to do with his "re-routed" server which just goes to NYC (which since it's literally right above Montreal is beyond me why he re-routes there) because I ordered that package thinking the same thing as above (OVH Canada is Northish but really it's in Montreal) and it turns out it was just a worst idea because from what I've experienced with Crident which is the only second OVH host I've used in the "US" it's a trumedus down spike in ping with them than GMC's re-routed locations (also my ISP never changed during this time, still got the same one).

Finally their "DDoS" protection on their website (which I'm supposing GarnetGaming is on their premium package) is really not protection tbh, I had a client who was talking to me and he said "Do you want to see me shut down GarnetGaming's website with a $5 booter?" I asked the host n' shit, he said GMC, so he ran the booter and bam, Error 500 until he stopped, I talked to etrug about it to notify him and he just said "It's most likely mybb's php" and I didn't know how to respond back to that (cuz it's pretty dumb considering it was just a generic skid ddos attack) so I let him have that and went on my way. (also I had a mybb forum on nfo web hosting and he was unable to take that down, unlike Garnet's)

#Crident's Web Hosting

I got a really nice trial from them it was basically their first time so I guess it could be considered as beta testing but me and josh decided to see what we could do, some settings where there that shouldn't have been able to be fittled with (that's fixed to this current date btw) and we tested the boundaries of a shit ton of requests to the server with a for loop and some php magic, after a while the server held up tightly, it was clear though the requests where succeeding after drizzy showed us the list but regardless without the limit (that is now in place and not tamper-able anymore) the web server held up against it all for a long time. After having some fun with that I never really noticed any issues they've had with their web hosting, the panel (cpanel) was delightful, speed was high and we never had any hardware issues (such as outta ram n' shit) so I think it's safe to say Crident Web Hosting is pretty stable to get at it's current stage.

ByLunaversity.me 🌸at 13th of November 2017 05:29:31

most of the gmod population is in the East.

OVH have launched their Virginia Data Center, it's in beta right now by the looks though. We'll hopefully have servers there soon. If you want, you can help get Crident near you by voting here: https://crident.com/global

ByLunaversity.me 🌸at 13th of November 2017 05:29:31

cause even though their support has been absolute dog shit, shared hosting gives better support than them, especially Crident.

I can't agree with people on OVH's support though, they're always awesome with me, and I seriously rep their support and company as a whole. We're looking to exclusively use OVH in Crident's growth plans for 2018, and I couldn't be happier with our choice.

As a general note, OVH's DDoS mitigation is really stepping up fast. They're currently phasing their "next-gen" mitigation in, and they've got some really cool shit going on which is just incredibly fast. With their new in-house mitigation, they're using programmable FPGA chips which are faster than CPU's for this use-case. I don't think any company can provide protection better than OVH for their price. I'm yet to see a successful attack on our OVH servers since the start of 2017...

would love to see Crident dip in the US. Their hosting would blow up. Many people having issues with gmc currently. We use them and we have decent support but are always told "hire a dev". Crident from what i hear has good support. GMC currently has been crashing a ton. Most servers i play on with gmc deal with the same issues.

By$$FUMONEY$$at 18th of December 2017 15:35:09

would love to see Crident dip in the US. Their hosting would blow up. Many people having issues with gmc currently. We use them and we have decent support but are always told "hire a dev". Crident from what i hear has good support. GMC currently has been crashing a ton. Most servers i play on with gmc deal with the same issues.

Crident's actually going to start selling servers in Virginia when OVH is done with them (which is very soon, even possibly next month as drizzy keeps saying).

And for the US West apparently they're already building the walls for the datacenter at OVH so expect that soon too.

No offense but GMC is just horrible. When I used them I always started capping 100% CPU at about 40 people and I was told to “hire a developer” and it was because I just “threw whatever scripts I could find on it”. I come to figure out it’s actually their hardware as I expected. They use an older E3 that clocks quite low and is horrible for gmod. Their machines also use HDDs, making load times and spawning severely slow and laggy. The only benefit was the ping but crident has been able to match that now.

I was told to “hire a developer” and it was because I just “threw whatever scripts I could find on it”

best tech support 2017

best

Bit late, but just to confirm - we plan to have Virginia (US East), Texas (US Central) & a Oregon (US West) location planned, too. Texas will be last - you should expect east & west coast to open in pretty close time to each other.

For you internationals, expect Germany after US-East & maybe Poland & Amsterdam in 2018, too.

Edited

By$$FUMONEY$$at 18th of December 2017 15:35:09

GMC currently has been crashing a ton. Most servers i play on with gmc deal with the same issues.

That's interesting - a community server I manage on GMC is crashing more than the server with Crident. I've been trying to work out why now for ages, but it's the first time I've heard someone say this is a GMC-specific issue. Cheers for the heads up - I'll speak to ertug to see if he might have any suggestions.

Edited

ByLunaversity.me/c 🌸at 18th of December 2017 18:37:09

And for the US West apparently they're already building the walls for the datacenter at OVH so expect that soon too.

That was actually a while back - I was just mentioning it at that time because of the trump wall meme value. US east & west are both in phases #2, entering phases #3 - which is the last phase of construction before some cloud-based testing, and then beta-server offerings which we're on the invite list for, and then it goes public.

The really exciting thing about the US side of things is that OVH can really start to leverage their dark fibre network they have spanning the US - which is also interconnected with their EU dark fibre network via private under-sea cables. It'll be interesting to see what kinda ping's we get when they start locking their routing down. At first you can expect it to be underwhelming, but as they get the routes more efficient and build their private network out more, I think we're really going to see some benefits in the US.

I've never had any issues with GMC, never went down, no lag, good ping. Support is limited to the service itself however, he won't solve other issues. Just my opinion

It's probably best if just I go ahead and clear up some claims and accusations.

By$$FUMONEY$$at 18th of December 2017 15:35:09

would love to see Crident dip in the US. Their hosting would blow up. Many people having issues with gmc currently. We use them and we have decent support but are always told "hire a dev". Crident from what i hear has good support. GMC currently has been crashing a ton. Most servers i play on with gmc deal with the same issues.

I need you to be as specific as possible with what you mean when you say that servers hosted through GMC crash a lot. I'm not aware of any issues at all with any of our locations or with any client regardless of how big or small where they are crashing without reason. Every single client who has experienced performance issues or had their servers crash was caused directly by either their server addons (lua) or in-game related (physics, collisions). This assumes that there was nothing on the network level which I am always monitoring. There has been a recent trend of physics-based server crashes a couple of week ago. Even many large servers experienced the issue regardless of their custom in-game countermeasures. Ticking the anti server crash option in the command line editor appeared to fix many of the exploits.

Keep in mind that we will not rip apart a clients server to find out what server add-ons or in-game related exploit are crashing their server or giving them performance issues. It should be obvious that optimization is all up to them and should not be a service offered for free by any hosting company.

BySkepticat 18th of December 2017 22:53:56

No offense but GMC is just horrible. When I used them I always started capping 100% CPU at about 40 people and I was told to “hire a developer” and it was because I just “threw whatever scripts I could find on it”. I come to figure out it’s actually their hardware as I expected. They use an older E3 that clocks quite low and is horrible for gmod. Their machines also use HDDs, making load times and spawning severely slow and laggy. The only benefit was the ping, but crident has been able to match that now.

If you were experiencing 100% CPU at only 40 players then there are add-ons that are ridiculously unoptimized on your server. The hardware and specifications that we use for our US East location are designed to give you a recommended player slot value of 64 players on a decent setup running DarkRP. The IO speed has nothing to do with how fast people load into the server, it all has to do with your server addons networking Lua to the client on sending client info.

When I build the new game server machines for Dallas, the hardware will be quite better than what the recommended player slot value is, however, you shouldn't have had issues to begin with unless there is something on your game server that is seriously wrong. My guess is that you would have hit a max of 75-80 players on hardware that is designed to hold 100 players (eg. premium package) which is also very low compared to what you're supposed to be able to hold on your servers. I would highly look into optimizing your server if you had not done already.

Good luck.

Bydrizzy.vip - cridentat 2nd of January 2018 12:06:31

That's interesting - a community server I manage on GMC is crashing more than the server with Crident. I've been trying to work out why now for ages, but it's the first time I've heard someone say this is a GMC-specific issue.Cheers for the heads up - I'll speak to ertug to see if he might have any suggestions.

I would indeed recommend contacting me directly for some suggestions. There would be anything that wouldn't directly be caused by either something in-game related (collisions, physics), server addon related (lua) or network related (I've heard that the routing from LON can be funky at times but it would not be a direct server crash).

I agree that the disk speed has nothing to do with load times. It’s just annoying when you have a server with cars or other large entities. I noticed when I switched off GMC to a host that used SSDs the load time for large entities was atleast 5x faster and practically was un noticeable when cached. When I was with GMC on the HDDs, the server would lag when spawning cars even after they were fully cached.

BySkepticat 2nd of January 2018 23:10:17

I agree that the disk speed has nothing to do with load times. It’s just annoying when you have a server with cars or other large entities. I noticed when I switched off GMC to a host that used SSDs the load time for large entities was atleast 5x faster and practically was un noticeable when cached. When I was with GMC on the HDDs, the server would lag when spawning cars even after they were fully cached.

The server will not lag when you spawn a car as long as you do not have an addon that hooks into vehicles. The only other factor is if you precache a model or not. That is technically a clientside thing. The only way that the server will lag when you spawn a car is if you have an addon such as a vcmod ** and or alternative ** that is very heavy and unoptimized. I can't say that I have ever seen people spawning a vehicle lag the server other than if it wasn't precached using Lua.

By[GMCHosting™] ertug2at 2nd of January 2018 23:29:20

The server will not lag when you spawn a car as long as you do not have an addon that hooks into vehicles. The only other factor is if you precache a model or not. That is technically a clientside thing. The only way that the server will lag when you spawn a car is if you have an addon such as a vcmod ** and or alternative ** that is very heavy and unoptimized. I can't say that I have ever seen people spawning a vehicle lag the server other than if it wasn't precached using Lua.

do u just lie for a living or what

I like how multiple people have complained about crashing servers etc and he is turning around and blaming addons etc even to people that have coding experience. You could just admit your servers hardware has issues.

ByQuack🦆at 2nd of January 2018 23:35:49

do u just lie for a living or what

I'm being as truthful as I can. I'm telling you exactly what I have experienced in my many years of doing this.

ByCommander Santaat 2nd of January 2018 23:47:00

I like how multiple people have complained about crashing servers etc and he is turning around and blaming addons etc even to people that have coding experience.You could just admit your servers hardware has issues.

I'm completely open to constructive criticism but they need to be as specific as possible. I'm not sure what coding experience has to do with seeing the same problems come up hundreds of times throughout time. This isn't my first rodeo.

gmc killed my mom

In the past few replies I've seen people blame Lua and engine crashes on the host. My disappointment is immeasurable and my day has been ruined. Competence is a neat thing when it comes to this but at least be able to recognize that regardless of the machine, srcds isn't going to just crash unless there's an engine error, a Lua panic/infinite loop, module crash, or the process runs out of memory.

Also GMC killed my whole family.

My brain has seized all functions after reading this... GMC works fine, move on with your life.

I think with disk speed more than entities where you notice the huge hit is when there are addons like ULX, ARCBank, and others that store a lot of data in text files.

Afaik, the game thread freezes while writing to the disk & I/O block could worsen this. It’s more measurable on shared hosting where typically the disks are under more load than say a private dedi

I can say that my US server (w/ gmc) crashes about 3 times for each time my EU server crashes & the EU (w/o gmc) server gets about double the player count. The servers are as identical as I can make them & I have actually asked you about it before Ertug.

I don’t know if its a real thing on your end - but I at least wouldn’t brush off the reports. I remember you mentioning to me that you have something on your servers to try and stop crazy physics crashes, I would look at that first personally, as physics is fucky

Edited

ByKingofBeastat 3rd of January 2018 01:29:39

In the past few replies I've seen people blame Lua and engine crashes on the host.

It is reasonable to consider the possibility something could be causing issues on a host level if a series of people claim so. It’s not impossible how the host manage their file system, could cause some freak crash issue for example; if he had something which updated file permissions while the server was running. This can trigger those files to lua refresh, and I could imagine every lua file suddenly reloding causing issues.

I don’t know or think this is the issue (if there is one) - I’m simply using it as an example that the possibility is there if you actually think.

Ertug has also told me (and others) that he has some module on his servers which aims to prevent crazy physics crashes. Isn’t it also possible this could induce a crash?

Edited

By[GMCHosting™] ertug2at 3rd of January 2018 00:20:09

I'm not sure what coding experience has to do with seeing the same problems come up hundreds of times throughout time.

I think he’s implying that by these people having this knowledge they’re more credible, and in turn less likely to simply be having a crash caused by an addon or something else trivial

But you say you’ve seen this issue hundreds of times? Isn’t that concerning?