Did my hosting provider respond appropriately?
April 15, 2013 6:01 PM Subscribe
My blog made the front page of a major social media site this morning and the provider promptly locked my entire account for "abuse". It took 6 hours for my blog to come back online. Was that a typical response?
My Wordpress blog is hosted by a fairly well-known (small-ish) hosting provider. The one time my blog experienced a huge spike in traffic and we exceeded our bandwidth, it was as simple as upgrading our hosting plan to a higher amount, and we had almost no downtime.
But this morning, when we were Slashdotted (or whatever you call it these days), our provider shut us down for "abuse" when the high volume of requests took up more load on our (shared) server than allocated/anticipated. (We had not come close to our bandwidth caps.) It took 2+ hours for the "abuse" team to get back to us with a response regarding our options, and more hours between the time I submitted a credit card payment for an upgrade to a cloud-based virtual dedicated software, and then time for their techs to help us through the migration process. All-in-all, a 6-hour process from the first report that we went down to when it was available again.
They were downright decent at the end there with providing assistance (and nameserver redirects while we worked things out); my kvetching at them via Twitter, their FB page and in comments to the post driving traffic to my blog may have had something to do with it.
I now realize that it was unreasonable for me to expect that the load spike would be expediently handled just like the last time. (This morning's spike was orders of magnitudes higher than my last spike.) And I also appreciate that I've been a small customer with almost nonexistent bandwidth and storage needs.
That said, was a 6-hour turnaround time reasonable? I was in "take my money, fix this" mode from my first contact with them, but they were unresponsive to my pleading to give me a timetable throughout the entire process.
My Wordpress blog is hosted by a fairly well-known (small-ish) hosting provider. The one time my blog experienced a huge spike in traffic and we exceeded our bandwidth, it was as simple as upgrading our hosting plan to a higher amount, and we had almost no downtime.
But this morning, when we were Slashdotted (or whatever you call it these days), our provider shut us down for "abuse" when the high volume of requests took up more load on our (shared) server than allocated/anticipated. (We had not come close to our bandwidth caps.) It took 2+ hours for the "abuse" team to get back to us with a response regarding our options, and more hours between the time I submitted a credit card payment for an upgrade to a cloud-based virtual dedicated software, and then time for their techs to help us through the migration process. All-in-all, a 6-hour process from the first report that we went down to when it was available again.
They were downright decent at the end there with providing assistance (and nameserver redirects while we worked things out); my kvetching at them via Twitter, their FB page and in comments to the post driving traffic to my blog may have had something to do with it.
I now realize that it was unreasonable for me to expect that the load spike would be expediently handled just like the last time. (This morning's spike was orders of magnitudes higher than my last spike.) And I also appreciate that I've been a small customer with almost nonexistent bandwidth and storage needs.
That said, was a 6-hour turnaround time reasonable? I was in "take my money, fix this" mode from my first contact with them, but they were unresponsive to my pleading to give me a timetable throughout the entire process.
Find a new provider. Seriously, unless you took down the server you were on, it is unacceptable to take down your site for exceeding bandwidth usage. You can't risk sticking with them.
posted by graymouser at 6:21 PM on April 15, 2013 [2 favorites]
posted by graymouser at 6:21 PM on April 15, 2013 [2 favorites]
Depending on the hosting provider's resources and the version (and features) of the Wordpress instance you were using, they may have been doing the best they could with what they had available. Given their size and history of response, it's up to you to decide whether you're comfortable keeping your business there or moving it to a provider better equipped for what you expect the world at large may throw your blog's way.
posted by thatdawnperson at 6:25 PM on April 15, 2013
posted by thatdawnperson at 6:25 PM on April 15, 2013
Is it appropriate? Absolutely not. Does it happen pretty often, especially with smaller, budget shared hosting providers? Yep.
About the only real way to make your feelings known (unless they are truly committing to improve their procedures already) is to find a new hosting provider and design scalability into your infrastructure. If Wordpress is the primary app you're focused on, wordpress.com is worth researching, as this stuff is their business.
posted by zachlipton at 6:25 PM on April 15, 2013 [2 favorites]
About the only real way to make your feelings known (unless they are truly committing to improve their procedures already) is to find a new hosting provider and design scalability into your infrastructure. If Wordpress is the primary app you're focused on, wordpress.com is worth researching, as this stuff is their business.
posted by zachlipton at 6:25 PM on April 15, 2013 [2 favorites]
It's not reasonable, no. But as people are saying, it happens with some providers. Get a new host.
It could be worse, mind you! I once had a site linked to from CNN and got a $3,000 bandwidth bill at the end of the month.
posted by Jairus at 6:28 PM on April 15, 2013 [2 favorites]
It could be worse, mind you! I once had a site linked to from CNN and got a $3,000 bandwidth bill at the end of the month.
posted by Jairus at 6:28 PM on April 15, 2013 [2 favorites]
Best answer: That they disabled it for "abuse" tells me they're dealing with this by the seat of their pants. I mean really, this has been a solved problem for a decade, typically by upgrading your bandwidth plan (sometimes temporarily). Experienced providers know that site popularity is a feather in their cap, but their actions here say to me that they don't want to host popular sites (because they took your site down). New provider.
posted by rhizome at 6:48 PM on April 15, 2013 [8 favorites]
posted by rhizome at 6:48 PM on April 15, 2013 [8 favorites]
Do you have a cache plugin enabled for WordPress? Super Cache or W3 Total Cache are what I've used in the past. A lot of the time it's the constant querying of the database that kills a site on shared hosting. With caching, you eliminate a lot of this - which really helps when you get Fireballed (aka slashdotted). I don't know that it would have helped you in this case, but it's generally a good idea, regardless.
I'd expect my host to do their best to keep my site online if I got a surge of traffic.
posted by backwards guitar at 7:09 PM on April 15, 2013 [1 favorite]
I'd expect my host to do their best to keep my site online if I got a surge of traffic.
posted by backwards guitar at 7:09 PM on April 15, 2013 [1 favorite]
If they took down your site because of "abuse," the problem probably wasn't in your server. You probably were getting more traffic than their router could handle, triggering their anti-DDoS safeguards, to keep everything they host from going down with your site. They did do everything in their power to keep your site (and the sites of all their other clients) up and running. All your money wouldn't help them if their bandwidth is saturated and they don't have another pipe (which most small hosts don't). And they can't give you an estimate of when it will come back, because they don't know when the spike will subside. The 6 hours was probably the amount of time it took for the traffic volume to drop below that threshold. For a smaller hosting provider, this is both understandable and reasonable. They're simply not in the business of hosting sites that get linked from major aggregators or news sites.
Imagine your hosting provider is a rubber boat. They've agreed to take on four passengers and their passengers belongings, with the mutual understanding that the passengers' belongings aren't going to consist of much. Then you come along and drop a giant old-school safe into the boat. If they can't somehow deflect the safe into the water (or at least bounce it off the side), the whole boat is going to capsize, along with all the other passengers. There is no way they can scale up the size of their boat quickly enough to catch your safe and stay afloat.
Even for a large hosting provider, scaling from zero to Slashdotted (or Reddited, or whatever) is likely to cause hiccups. The difference is that a large service provider has somewhere to expand to if there's an emergency like that. To overextend my analogy: they've got a yacht, and they just have to make sure the deck is strong enough to hold your safe.
The question is, do you expect to get linked again, or to be able to convert this spike into a sustained higher level of traffic? If you're expecting another spike, you may want to find a new host. Even if you're not, you can probably find a host that's large enough to be able to absorb a massive spike without collapsing. But you don't need to run away screaming.
posted by smoq at 7:42 PM on April 15, 2013 [1 favorite]
Imagine your hosting provider is a rubber boat. They've agreed to take on four passengers and their passengers belongings, with the mutual understanding that the passengers' belongings aren't going to consist of much. Then you come along and drop a giant old-school safe into the boat. If they can't somehow deflect the safe into the water (or at least bounce it off the side), the whole boat is going to capsize, along with all the other passengers. There is no way they can scale up the size of their boat quickly enough to catch your safe and stay afloat.
Even for a large hosting provider, scaling from zero to Slashdotted (or Reddited, or whatever) is likely to cause hiccups. The difference is that a large service provider has somewhere to expand to if there's an emergency like that. To overextend my analogy: they've got a yacht, and they just have to make sure the deck is strong enough to hold your safe.
The question is, do you expect to get linked again, or to be able to convert this spike into a sustained higher level of traffic? If you're expecting another spike, you may want to find a new host. Even if you're not, you can probably find a host that's large enough to be able to absorb a massive spike without collapsing. But you don't need to run away screaming.
posted by smoq at 7:42 PM on April 15, 2013 [1 favorite]
Response by poster: Thanks, all, for the gut-check.
The "abuse" they cited was that on our shared host, we were taking up 40% of the CPU load and were single-handedly causing Apache to exceed whatever max processes limit was set. This makes sense to me, and they certainly seem to have hosting plans that could handle the greater load. They migrated us to a VPS host which is very zippy, it seems.
But what was confusing to me was the 4+ hour delay from me slapping a credit card down to being back online on the VPS host. And I've been trying to figure out if I've missed something. The responses here helped me get more comfortable with my own assessment of things. Thanks!
posted by QuantumMeruit at 7:49 PM on April 15, 2013
The "abuse" they cited was that on our shared host, we were taking up 40% of the CPU load and were single-handedly causing Apache to exceed whatever max processes limit was set. This makes sense to me, and they certainly seem to have hosting plans that could handle the greater load. They migrated us to a VPS host which is very zippy, it seems.
But what was confusing to me was the 4+ hour delay from me slapping a credit card down to being back online on the VPS host. And I've been trying to figure out if I've missed something. The responses here helped me get more comfortable with my own assessment of things. Thanks!
posted by QuantumMeruit at 7:49 PM on April 15, 2013
If that's what they said, and that's what they did, then I take back what I said, and yeah there's no good reason it should've been down for that long.
posted by smoq at 8:27 PM on April 15, 2013
posted by smoq at 8:27 PM on April 15, 2013
They're bad at communicating. You might get a better response if you call them.
I second backwards guitar's caching recommendation - I like WP Super Cache, personally.
There are other things you can do on a VPS that you can't do with shared hosting, but you'll get a pretty big boost out of setting up caching first.
posted by Pronoiac at 2:20 AM on April 16, 2013
I second backwards guitar's caching recommendation - I like WP Super Cache, personally.
There are other things you can do on a VPS that you can't do with shared hosting, but you'll get a pretty big boost out of setting up caching first.
posted by Pronoiac at 2:20 AM on April 16, 2013
This is not an appropriate response by the host. Could you send an e-mail to Matt Mullenweg (project lead for WordPress) and tell him who the host was and what happened? We have contacts at a lot of hosts, and can lean on them when stuff like this happens.
posted by markjaquith at 2:27 PM on April 16, 2013 [1 favorite]
posted by markjaquith at 2:27 PM on April 16, 2013 [1 favorite]
This thread is closed to new comments.
posted by chrchr at 6:17 PM on April 15, 2013 [12 favorites]