I would like to know as to what amount of expectations should one have when opting for a VPS.
i.e. what's the ideal kind of website that will work fine with a VPS ideal site in the sense that ... the bandwidth and resources (or other criteria) the site consumes.
The term Managed Services concerns me with the way it is tossed around in the industry and I am curious to see what other hosts refer to as managed services, how they implement them and also what the consumer expects to receive on a managed server.
In going over one websites description earlier today they had more than five pages devoted to managed services and in my opinion said nothing. In fact at one point they mentioned their backup power, which while I will agree with that it is very important, that is infrastructure and IMHO has nothing to do with backup services.
Anyway I am interested in your view and definition of what you expect to receive on a managed server.
I'm experienced with various flavors of Unix but haven't taken the VPS plunge yet. My main fear is that I'm going to end up spending a lot of time managing the server, especially security. I'm leaning heavily towards a hoster that's well regarded on this forum but was surprised when I learned the default managed VPS they deliver is fully open. It also sounded like it was mostly up to me to do the hardening. I've seen the really helpful "HOW TO: Secure and Optimize your VPS" article and I'm okay carrying out these steps initially but all of this makes me wonder:
1) Shouldn't a brand-new managed VPS come completely closed and leave it up to the user to open the few services they need?
2) How much security work should I expect a managed hoster to provide up front and on an ongoing basis?
3) How much security work should I expect to put in initially and ongoing to keep this site running smoothly and securely?
Don't get me wrong, I'm not expecting this to be "fire and forget" but I don't want running this server to become my day job. And after all, with my current shared hosting package I don't have to do *anything* - other than the occasional outage, it just runs.
moving my colocation (just a single 1U server) to another provider. Is it reasonable to expect to be able to pay the original provider a fee to ship the server to the new location for me instead of requiring me to fly to the DC to pick up the server and do it myself?
View 2 Replies View RelatedThis is my review of my initial setup with Pacific Rack.
After the issues with the fire at TP, I decided to pickup another server at a different data center as an additional server.
I ordered a Supermicro, which is a quad core processor. The price was well within the range, the extras were all within range as well, except cpanel, which was $35.00. I'd rather see that at $25.00, but overall that was fine. Total price as $266, right around what I expected to pay.
Setup took a few days, right at around what I expected.
When it was delivered, I asked why I couldn't log into WHM. They told me everything was fine. After some communication and about an hour and a half they said that they forgot to install cpanel... That wasn't a good start.
They failed to meet expectations because of the following:
1. RDNS was not setup at all. I shouldn't have to request it be setup. This is small, but if I didn't know to check/ask, it would be a major issue in the future.
2. The box always fails to reboot and requires manual intervention. Why can't they fix that? or even address the issue?
3. cPanel was not installed properly and when I asked for help, they told me that the box was unmanaged and I choose the configuration. If it didn't work right out of the box, the fault was my own for choosing that configuration.
Alex said, "You have received a standard Fedora 64 bit install, that we downloaded just for you. What you need must not be included in the standard install, or the cPanel install.
If you want us to manage your server, then you will have to subscribe to our server management service which is $29/month with a 6 month commitment. Otherwise you can just google for the command to install an RPM, as I don't know if off hand, but I know its really simple."
Just to comment, if it turned out to be as simple as installing an RPM, then I would have just killed myself due to the shame of it.
Turns out after working with cPanel support, it was an install error, cPanel was not installed correctly. (Don't know if that is PacificRack's fault or cPanel's, but I know it's NOT my fault and I would expect the data center to get my server up and running properly so I could manage it)
My expectation was that I would be delivered a box that worked properly and if it didn't, PacificRack would help me get it working so that I could use it. It appears that PacificRack's expectation was they just install the crap, it's up to me to make it work. If the install is bad, I can just Google it.
I'm fine with managing a server, I have four years experience running serveral servers, but I am not able to debug install issues where the box NEVER worked properly in the first place.
So, in this area, they fail to meet expectations.
I also found their technical staff to be arrogant and unfriendly, though I'm personally arrogant and unfriendly, but I always am very nice to support people, no need to cut my own throat.
Overall, I'd say that my experience doesn't seem to be typical, but the fact that they were unwilling to resolve the issues really bothers me. I detest companies who hide behind "unmanaged" as a shield against providing the services that they should be providing as a data center. I also refuse to pay a $29.00 a monthly fee to get a managed server (basically a $180.00 commitment) to get them to deliver a working server, when that should be a "Free" service provided when I purchase.
The DNS servers are pacficrack work well and the connections are fast and reliable.
The server still is not working up to expectations and their support people seem to be unwilling to get it to meet expectations with out going "managed". Now, honestly, $180 would be far cheaper than any other options, but its the principle of it.
Overall, I'd rate pacificrack about average. The server is a good server and at a reasonable price. The support people where knowledgeable. The turn around on tickets was very good, no ticket sat for more than a few hours. It really seems like pacificrack is a very small operation. I think that is a big plus.
At this point, I'm not sure if I should just move on or continue to try and get this server to work properly. I'll make a decision in a week when the server is close to renewal.
I had three VMs with Fsck VPS, dating back to before they got hacked in June. I've been paying the bill since then, I imagined as a kind of insurance, so that I had the VMs handy if I needed to use them in a big hurry. Last week, I tried to log in, and found that my three VMs didn't exist, anymore. As far as I can tell, the VMs haven't existed since the June break-in.
SETUP / PROVISIONING
My first reaction was "They've been billing me for three months, and providing nothing?!??!" I'll be honest, I was pretty tweaked, but after I'd calmed down I decided to see how they handled the situation. So I submitted a ticket asking for an explanation: How long had the machines been down for, and what would it take to get them back up and running?
It took about a day, but we eventually established that VAServ could build three new OpenVZ VMs, and that they would give me three months' credit for those three machines. Since I actually do need the VMs, and I didn't really want to fight about the billing, I decided to go for it. It took another day, but I did get three new machines up and running.
Unfortunately, I do have some complaints about the process, specifically:
- VAServ's technical support is very inconsistent, and different techs seem to have vastly different levels of communications skill and professionality.
- Many of the techs don't seem to bother reading your ticket, beyond the subject. They tend to only be capable of answering the first question in each ticket/email, and they ignore anything else you've asked.
- After the FSCKVPS/VAServ buyout, following the break-in, the HyperVM control panel was disabled. If you need a reboot, or a root password reset, or anything that you can't accomplish yourself by SSHing into the VM, you have to open a ticket. (Seems like a chancy proposition, now, to me.)
REBOOT-AND-PRAY
Today, I started seeing memory allocation errors in running programs. The machine mostly worked, but certain operations (shell scripts, in particular) would error out. I opened a ticket asking for some guidance, and within less than 10 minutes, the VM started rebooting. I got an update about the ticket a few minutes later, and was told that the VM had been reconfigured (increased memory allocation limit) and rebooted.
I was pretty mad about the no-notice reboot. I'd been in the middle of editing a bunch of configuration files, and I lost an hour of work. It just seems so unprofessional and inconsiderate for VAServ's technician to bounce the VM without confirming it with me, first.
I did get an explanation/apology from the tech who rebooted the machine. I asked him to have his supervisor contact me, which took a few hours, but I did hear back. The supervisor wrote:
"...we reboot the vps if we found any VPS out of memory. Normally most of the service stop working or access got killed when VPS is out of memory..."
To me, it sounds like the reboot is a standard procedure for a common problem. Given that kind of environment, it's only natural that the tech's first impulse would be to reboot, given a ticket about memory errors.
At the same time, it's also indicative of a bottom-of-the-barrel service, isn't it?
- Memory problems seem to be common--is that because they're over-subscribed? Does your 512MB allocation mean anything, or is it just talk?
- The staff can't / won't bother to read through a ticket and give it some consideration.
- The staff has an itchy reboot finger. Their first impulse is to power-cycle, rather than to try to understand and fix the issue directly.
FOR THE FUTURE
I do intend to continue using VAServ / FSCKVPS, at least for now. They're really cheap, about $10/month for a 512MB VM, and I can mostly get done with what I need to do. But this is a qualified opinion. I am solely using these VMs for simple R&D projects: Quasi-professional work, stuff that nobody is currently paying me to do.
Given my experiences so far, I would never trust these guys with a real, money-making business project. VAServ / FSCKVPS is suitable for toying around with, or if you're flat broke, but I wouldn't bet my job on them if I could possibly help it.
I'm setting a calendar reminder for myself, right now, to check back in another month or so with an update to this post. Assuming I'm still chugging along with these VMs, I'm going to make a point of posting my impressions on a regular basis.
While dedicated server providers are quick to promote that they support the server hardware how proactive do you expect them to be in terms of identifying hardware problems?
When hardware goes bad a customer might see performance problems and report it, or perhaps notice hardware errors in the logs.
Or is it the host's job to be monitoring hardware components status? Would you consider such monitoring to be part of the basic service or an optional management extra?
I'm interested in your expectations with a typical dedicated server.