New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Should I switch to a dedicated? Or when to switch?
Currently, I have something along the lines of 9 VPSes which together have 14GB of RAM and 1TB storage (Less than 100GB are SSD). Everything averages to $185/year which is $15/monthly on average.
Like some other people here, I simply couldn't pass a few offers, mainly because I have a few stuff I like to run into separate systems and I can't install VMs in most VPSes, even if I could, I would need one bigger VPS to allocate everything else. So I thought about getting a dedi instead and making multiple VMs to host the various services I'd use.
I saw a dedi on DRServer for close this value, so I was wondering you guys opnion (and reasoning) on whether I should switch to a dedi or keep running multiple VPSes.
Thank you.
Comments
It really comes down to the question, 'are you comfortable having all your eggs in one basket?'. If you are then managing one server will, of course, be more efficient and less time-consuming.
I prefer to have a number of VPS at different providers for various uses so that I can be reasonably assured they won't all go down at the same time.
Really depends on your use case and how comfortable you are consolidating.
what Lee ment was actually the 🍆
on a serious note, better to distribute your stack and backup.
Well, I do think it would be easy to maintain and support. I only run services for myself and a few family members (less than 10 overall users) and I plan using something else to backup. As it is I'd mind having an uptime of 99.7% (that's like less than 2 days per year) overall on my services.
I've never managed so many servers at once, so it's getting kinda overwhelming. I do plan on being a SysAdmin on the future, but I'm not* there yet (although I'm still managing everything just fine), that's why I thought having a dedi would be better.
Asked for opnions mostly because people here are way more wiser on the subject than am I.
I sort of agree with @Lee, on the flip side you could ask "are the VPS provider's economy of scale on the whole better than a dedicated server provider's"
Personally, I think a dedi would be the way to go. Assuming you have backups and a contingency, worst case scenario is you have to use them. Theoretically, less hassle using one provider, one invoice - but also a single point of failure. For simple sites this should really be non-issue.
VPS's only scale because some customers use less resources than their fair share. In the end, they're using hardware allocated to a dedicated server. There are so many unknown variables trying to answer this question, in some ways it's not worth asking.
With the VPS provider's you're playing roulette with their ability to allocate resources fairly and have a valid long-term plan. If you're going with larger providers it's perhaps less of a problem.
Ultimately if you command a decent wage per hour, you're better off fire and forgetting, a dedi, with known resources, with a well-known provider. Alternatively you can choose low end providers and find yourself switching providers-per-site every 6/12 months. Even with semi-respectable providers here who can manage their resources - you might find them migrating from rented IPs to their own range and then require you to change IPs - something that'll take your time.
IMHO the larger providers, dedicated providers provide you with more surety, but it's swings and roundabouts.
In the end, your time and effort is the major factor.
^^^ I consider this to be an important point since, imho, a dedicated server would give you the chance to improve your sysadmin skills. For instance, you will have a good time configuring a PROXMOX node and spinning some LXC containers to isolate and organize your services and your family's.
I do have a VPS (and other storage options) prepared to serve as a backup. In case everything goes down for a while I could find my way around.
Yeah, I was really thinking about doing just that, it's not rare a situation where I fuck installing something in a server and just have to start over, that's why I have different server for different purposes. But I think I will wait until BF to see there are good deals for dedis as most of my stuff was bought between April and June and a few are yearly.
Something to keep in mind: There often is a need or at least a desire for different IPs (and networks). Some TLDs for example require min. 2 name servers in different /24. But it's also desirable for reasons of availability.
On the other hand IPs are cheaper than VMs, so except for the above consideration putting all those VMs on one dedi is probably attractive. If your provider can give you IPs from different ranges (and you know how to configure networking in your dedi) that would probably be the best solution.
Another point to consider is that, depending on the performance of your VPSs a cheap dedi probably doesn't have the power to replace (or run) them all, especially not in the $15/mo range.
My final point is to keep backup in mind and to find a smart construct that would enable you to at least have some minimal service in case your dedi ever fails. So I'd suggest to keep or get one VPS with reasonable power (say 2 vCores) and a big fat reliable disk (preferrably Raided) as backup box that can also serve as an "emergency server", say for your most important web site and as secondary name server.
But frankly, with that budget of 15$/mo I'd rather look for a decent dedicated cores VPS plus a cheap one with a fat disk.
You touched on a lot of points that I didn't understand fully. For instance, I thought I could run Proxmox using only one IPv4 and then forwarding ports and putting it behind Nginx so I could customize which (sub)domain would connect to which service.
The only dedi I saw in that price range as of now is DRS one with an Atom C2750 with 8 cores @ 2.4
If I'm not mistaken, besides the one VZ, most of my CPUs are Xeon E5 (I would have to check, but a lot of them are 2680 v2). Except for two VPSes which have 2 vCPU, everything else has only 1 vCPU.
I was wondering that maybe in BF I could find a good dedi in that range.
For backups, I just got a 3TB HostHatch Storage server, I might as well run the main "fallback server" there.
I actually could use a dedicated cores VPS, but I was mainly looking into learning about Proxmox whilst keeping a separation of services
@Ouji
I don't know proxmox well but principally you can forward ports and play those games (but be aware that a fw uses resources too). Just keep in mind that many protocols use de facto standard ports. So, for example for web sites I'd suggest to stay away from redirection games and to just use virtual hosts (as seen and resolved by the http server) on 1 or 2 IPs.
As for a C2750, yes that's roughly equivalent to 8 cVores in VPSs and not a bad choice but there are caveats. You should for example have a good look at the flags that processor supports. But generally speaking and assuming that you don't run heavy duty stuff like much frequented web sites or need serious back end power (e.g. for a serious DB server) I think you should be fine with that kind of processor. But I'd still suggest to have at least 1 or 2 additional IPs; that would make some things easier, avoid some problems, and open some doors to interesting things to do. Oh and if any possible have IPs from 2 (or more) different /24s, preferably even from different networks (that saved my day more than once).
As for your existing storage server I think your plan is OK if the system isn't weak (~ if it has 2 vCores and at least 1 GB memory). The decisive point btw. is less the CPU type (e.g. e26xx) but rather how much of a real core a vCore equates to.
One last piece of advice: If you are experienced with linux you might want to use a small distro. It must not be as extreme as using e.g. Alpine or Tinycore but I'd advise against e.g. Ubuntu. After all with the kind of dedi you'll get for 15$/mo you don't have resources to waste.
And be absolutely sure to only get a dedi with some kind of remote console (IPMI, KVM, something like that) unless you are really very experienced. Funny panels and other gadgets are nice but a remote console that works even e.g. during booting the dedi is an absolute must have. Among other things that also allows you to install your OS choice rather than being limited to what a provider has templated.
@jsg
A lot of very good advices. Adressing a few points:
I won't host any heavy website or db. Just normal stuff, like file storage and a few webapps (most are not resource heavy). The only thing I might use that might drag a few resources is a Plex Server, but I didn't decide yet and it would be most of the time one stream at FHD or maximum two (very unlikely).
Again, I will probably wait until BF to see if a better dedi comes up. I will try to stick to my budget, only going out of it for a few extra IPs. But I need to see clearly the use case (besides failover).
The backup server has 3GB and 1vCore @ 3Ghz only, but I'm not entirely sure it is a fully dedicated core.
For distros I do like using Debian, I only run Ubuntu on boxes that needs a specific application I didn't get to run on debian.
Yes, I'm planning on renting only if it comes with a good remote console, as I might wanna install a custom iso (or simply install the system my way).
Start with hetzner SB. Get your hand dirty and feel more of it.
Having dedi doesn't mean you gonna save cost but gonna get more resources if you strike a good deal.
@mrclown Maybe they have better prices during BF or nothing change for the auction?
I know your comment touched on it, but if you're running your own name servers, definitely ensure the second server is on a separate physical server on a separate network. The entire point of having at least two name servers is in case one of them goes down. I think the relevant RFCs suggest a minimum of three name servers.
$185/year / 9 VPS = 20,5$ year for VPS
Time to move, my opinion, to a dedicated. Nothing different in management than a vps (look more carefully at the firewall) and to use a VPS (not low end) for backup.
IDK your providers but i tested time ago there, for fun, a few let vps providers (sry, no name, no shame, i can tell only i choosed several 20-40 bucks year vps) and their reliability (uptime, cpu processor, overcrowding, noisy neighbours) was horrible 100% (example: single vps with idling wp, just the sitemap plugin, generate 404 in sitemap.xml = you oversold like crazy that vps and you think customers are legit monkey).
Please look for a dedi if you look for reliability, bandwidth, no closed ports (a lot of low end vps close standard doors, maybe they think everyone is a
sysadminsysmonkey) and uptime.Even a basic hetzner dedi is better than 9 low end yearly vps.
Funny. My experience with LET providers was largely normal to positive.
No, not necessarily. 9 VPS come with 9 IPs, a cheap dedi with 1, and a dedi is bound to not be geo-diverse but 9 VPS (highly likely) are.
That is not to say that one should prefer VPS over dedi but that one should select properly. Like everything both dedis and VPSs have advantages and disadvantages and the use case may make one or the other the better choice.
Are you sure you're talking about LowEndTalk (LET) and not about LowEndBox (LEB)? LEB hosts are generally just a total shitshow and should be avoided.
I'm talking about LET (even the "404" shameless provider).
I've had issue with one of my providers only. The rest has been working flawlessly. I'm still taking all points into consideration and cooking the ideia, but I'm leaning towards a dedicated. I just don't know yet. I wish I knew if BF was gonna bring good dedi deals (because I know that for VPSes, it will).