All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
To what extent do you think about risks of hardware failure with various hosts/types of hosts?
I'm looking for hosting for a new client. The client is higher profile/higher visibility than many that I've had, and so I want to put them in a situation where I find the right balance of price/performance/reliability/resiliency for their site. In this case, price is probably lowest priority (I don't need to find a $3/month VPS, but also not looking for a $250/month dedicated server).
I've had really good experiences with a variety of hosting providers over the past handful of years (Linode, DO, NexusBytes, etc). Limited unplanned downtime/outages. But as I think about my new client, I'm starting to think more about the best fit for them. From a performance standpoint, I think a ~$100 dedicated box would work really well for their needs. Plenty of power and resources for their site. But then I start to worry about reliability. This is not a situation where 5 nines of uptime are required, but I also realize that I'd be at the mercy of the provider if a drive dies or some other hardware issue occurs. For providers that don't proactively monitor hardware health (and notify customers), that falls to me as the user, so that's an additional concern if those alerts come in while I'm sleeping, etc.
With Linode/DO/similar, they directly own their hardware and I expect outages to be minimal (they're not relying on a 3rd party remote hands in a datacenter to replace a drive, etc). If I rent a $100 dedi from a decent provider, I'm at their mercy if something happens.
I realize there are solutions for adding a bit of resiliency by doing things at the DNS level (auto failover, etc), but not all clients are using DNS providers that have those sort of capabilities built in (and aren't willing to switch to one).
What's the point for you when you decide that it's important to go with a provider that abstracts their offerings away from being tied to a specific piece of hardware (true cloud hosting)? I understand that at some level for every host, there's a hardware layer that has components that can fail. Even at AWS. Hardware issues could still occur with a cloud provider (fire in datacenter, flooding, etc) and cause outages. I'm less worried about those catastrophic outages (I have a redundant data backup solution in place and could quickly spin up a site on a different provider) and more thinking about those instances where a piece of hardware is down for 3 hours, etc.
**All that to say **-- how do you weigh the cost/benefit of being tied to a particular piece of hardware vs going with a provider that has more resiliency built into their infrastructure? Or do you just use the $5-$15 VPS that any provider offers and deal with the couple hours downtime when/if it happens once or twice a year?
Comments
Why dont you use a reseller hosting instead of vps?
I would never put a clients data on a LET server.
I use them for testing, small unimportant services. Testing before rolling something bigger out. I have a list of vendors I use for client data, the benefits, and the tradeoffs, depending on their budget. Failover, backup and redundancy is a must on my side but I prefer hosts to have their own backups as well.
I would use a LET server as a backup... But even then, I'm not sure I'd trust it as the only backup.
Just because you own hardware, doesn't mean you have people on-site 24/7, and thus remote hands provided by the DC might still very well be used.
If you can’t design your application around load balancing or automatic failover, then you need to go with a provider that has such huge economy of scale that hardware replacement will be quick and painless without human intervention
The only good options IMO are true cloud ideally with network boot drive, or dedicated server with a provider that does automatic hardware replacement (eg OVH)
Hardware replacement with OVH is not automatic or instant, can still take well over an hour. The good thing is that when you want failing hardware replaced at a scheduled time in the near future, they always do it exactly on time.
What you need is a high-availability setup with failover and backups. This can be done with any kind of low end servers. If you can't do this, then you might want to consider something similar to clouvider's vps. I believe they had some kind of failover system so you don't have to worry about hardware failure.
because typically they don't have the power/resources for the projects I need. And that doesn't offer any additional protection against hardware failure. Just a nicer admin interface to partition off client hosting accounts.
Yeah, I don't put any client sites on any provider that I haven't personally vetted/tested. I do make use of some LET providers storage servers as some of my backup storage. But then I use a big cloud provider for additional redundancy on backups.
Most of my client sites are Wordpress at this point. I know there are ways to do load balancing/failover at the hosting level, but there always seem to be tradeoffs that most of my clients aren't willing to pay for.
thanks...this sounds like a good option as well. But I'm likely looking for something available in North America, preferably US East coast. I've had good experiences with NYC-area and northern VA for my location/client locations.
If an OVH server stops responding to ping they will replace hardware if needed, without any tickets. Very few other dedi providers offer this
True cloud is the way to go of course if money is no object, but dedi with effortless hardware replacement beats most other options
I feel like the sentiment I've heard here at LET is that OVH support is not great. Other than when hardware is not working, I tend not to need support from my hosting providers. If OVH is so good at addressing hardware issues, perhaps they'd be a good option for me to consider. Their Beauharnois datacenter would be the logical choice for me--a bit further away than NYC from me, but not terrible. Worth considering, I suppose.
@muddy can't go wrong with Linode, DO or Vultr honestly. At this point these middle tier "cloud" providers all have some form of redundancy in storage in most cases. Just make sure you have the backup service they offer turned on (snapshots daily). If anything goes wrong and you are down, simply pop a new instance from your snapshots. Also keep other forms of backups, like every two hours, hourly, etc, you can do this with JetBackup to an external destination. If your client is high profile is ok to use those, but the strategy is to just use the top tier hosting providers, the one that the big boys usually use and pay a big monthly rate for it. I.E: Azure, AWS, Google Cloud. If for some reason an entire region is down, at least you get to say "yeah well Microsoft is down" or "Amazon is down".
From another pov, you can use any sort of middle tier, top notch provider and just have redundancy built in your system. Doesn't matter what provider you go with, if you dont setup redundancy it will eventually have downtime.
Vultr does not use any raid in their systems. So backups are super important