We're setting up a new business of Remote Data Backup service provision and our clients will range all the way from home users to large corporates.

We're starting small and hoping to grow from there and for this reason, we have taken up a 1U rack space at a colocation center in Toronto where our server will be housed.

Knowing the stability and reliability of Linux in the server environment, what would be the most suitable Linux OS to go for in this, plus, are there any 1U servers that come ready with Linux?

Also, what would be the most suitable software to run from the server-side point-of-view which would also come with it's respective client-side software which our clients would use to access their a/c's on our server and backup/restore their data as necessary? (something that would offer SSH to connect clients to us would be a gem)

You've already rented a 1U space while not having any idea if that is enough? How are you going to handle disaster recovery? Does the site you have chosen have redundant power circuits, Internet connections? It just boggles my mind that you are starting a business that you don't seem to know anything about. Do you have a business model created? I'm not trying to be cruel, but I think you need to do a lot more planning.

That being said, I assume you want to pay for support? Then Red Hat Enterprise Linux is the market leader in North America. Both HP and Sun/Oracle have some very good 1U servers and great support.

1 members found this post helpful.

Thanks so much, Bradvan and I have to say soory for a typing error on my part in my original post. What I was to write was, "we are taking up" instead of the "we have taken up".

As for disaster recovery we're considering having a backup server of our own at the same datacenter. The datacenter we're looking at offers all professional services including red. power circuits, bandwidth, etc..
We want to really start pretty small (not targeting corporates in the beginning) so I guess a 1U would be sufficient with about 500GB? (sorry if I sound silly here)

It's now the hardware and especially software part of things that I'm really thinking hard about and looking at the different options available. I would expect "alot" from the server-side software, while I'd be more than happy if we could get something on the GPL side and as our business expands, we can keep growing into paying for everything.

Well, you need to worry about what happens if this building burns, blows up, etc. If all of your data is at that one location, your customers won't be very happy if that happens. So, keep that in mind.

If you don't want to pay for support yet, I'd go with centos. It is re-packaged RHEL. So, it shouldn't be to odd to switch to RHEL when you are ready.

For hardware, take a look at the HP DL360 G7. You can get 1 or 2 processors with 2, 4, or 6 cores each. Up to 192GB of ram. It has 2 1 GB Ethernet ports and ILO3 for offline support. You can have up to 8 hot pluggable hard drives with a RAID controller. Dual redundant power supplies (you want each on a seperate circuit at the ISP). A very nice machine.

1 members found this post helpful.

I think you have a lot more to think about before worrying about the physical server side of things. What will you be backing up, and what is the OS your clients will be using? How will you get that data from the client PC/server/etc. to your server (you do plan to have this encrypted in some way while being sent over the wire, right?)? How long will you keep the backup for, will this be a set policy or is it configurable per client? The list goes on, this is something I have spent a fair amount of time trying to work out myself.

As far as the hardware to use I will say it depends on what OS you settle on and what type of management access you will end up having to the server. HP makes a very nice product with a mature remote management solution (iLO), but I would only use this if I had VPN access for this on a private network. I wouldn't want to have any OOB management interface exposed directly to the public internet. Another big plus for an HP (or Sun/IBM/Dell for that matter) would be hardware monitoring capabilities. HP has utilities which allow alerts to be sent when a drive fails, fan fails, etc. but these tools are not supported on every OS. So basically I would shoot to use an HP server, but ultimately if on a budget and I wouldn't have the secure OOB access I wanted or run an OS that is supported by the hardware monitoring utilities, I would go for something cheaper like SuperMicro.

I would also want to max out disk capacity since this is a backup solution after all. Personally I would likely do a small dedicated mirror for the OS and application level, and a larger array of some sort for the actual data.

One idea I have been spending a fair amount of time trying to work out revolves around using Solaris and ZFS. The thing about this that keeps me intrigued is the deduplication (not in mainstream release of Solaris yet) which would help get more bang for your buck on storage. It would also allow you to easily add storage to the pool and expand as needed. I have not been able to work out a good client end piece yet which would allow encryption, pass the traffic over the internet easily, work with Windows, and give end user at least some control over backup/restore.

1 members found this post helpful.

I have to agree. An awful lot needs to be considered. As Erik pointed out, who will be your clients. In my experience, most home users are willing to pay for a back up service. Most small businesses don't have the budget, and a large business would laugh you out of the room if you told them you had a single 1U server at a single co-location facility to back up their data. You need to seriously evaluate your clients. How many and how much are they likely to pay?

1 members found this post helpful.

Thanks so much, Bradvan and ComputerErik for your help.
We're not going to be placing both servers at the same location - yes, the same colo provider, but at their two different locations. Plus, I'm not placing the mirrored server in the beginning - that will come 30-60 days after the installation of the main server.

My initial plan is to start (very) small and in the beginning only have residential/small business clients who will be backing up their data with us and to grow from there. The is no intention of having large corporates as our clients for now. Pricing we would offer on the storage space plans - about $20 for 250MB (smallest plan). I really want to keep it lower, but the OEM software charge per user license.

The OEM software we're evaluating provides a server side as well as client-side solution, including an SSH based connection from client to server. Client side software is multi-OS compatible.

I cannot say how often each client will be backing up their data or the estimated sizes of each backup - we'll be providing each client with fixed maximum sizes and they can backup to their convenience - each client would have 24/7 secured access to their respective a/c's.
Thank's for this - it's something you've brought to my attention.
Thanks, I'm looking at this and also at the Intel SR1530CLR.
An interesting development came when I was googling and the QNAP Rack Mountable NAS server came to my view http://www.qnap.com/pro_detail_feature.asp?p_id=152
Would the Dedicated (rack mountable) NAS server be a good alternative?

Coming back to the HP, Intel, Cisco, SUN, etc...., what would be the basic configuration I can start with - my aim is to start really low on config and costs and to change the whole hardware within 6mths to a much more powerful setup.

One thing's for sure - where the server OS is concerned, my priority is Linux.

$20 for 250MB? Is that a monthly charge? What about more than 250MB? As far as I can tell if it is a monthly charge it is extremely high, and if it is a one time charge you will be losing money. Look at some of the competition you will be facing in the area, and you either need to offer better pricing (virtually impossible as a startup) or more/better services.

What software are you planning to use to accomplish the backups, and why is it necessary to use this package? If the user license is a one time fee for you it might be necessary to eat that fee to get licensed and then make up over time in a monthly fee.

Something else you will need to do is figure out how much usable storage you can get into that 1U. Since this is backup, and over the internet I would use SATA to keep prices lower. Settle on a few server models and configurations, then work out your cost to get that setup. You should then work out what you will need to charge in order to make the a profitable exercise.

Going with a NAS or SAN solution is an excellent idea, but will break the 1U form factor you originally were looking for. Let's say you can get 1TB usable space for client backups in the 1U space, what happens when you start using close to that 1TB? If you setup a NAS/SAN backend and the 1U server is just running your software it is easy enough (or should be) to expand that storage transparently to the application.

The NAS server isn't anything special. It is just a computer with an operating system that they have set up for you. You can do it yourself, or let them do it. It comes down to cost, security, and trust.

I've been puzzled by this thread since it started, but haven't responded, because I didn't want to be rude. However, it seems that you don't have any kind of idea differentiater -- that is, something that sets you off from the competition, gives you an edge, or is new and/or different; you don't have the detailed technical concept (you're asking here for software ideas); and you don't have a business plan (detailed costs of start up and operation, expected returns, time to break even, all with analysis and justification). All three of those are critical to launching a business.

If there were no one in this sort of business already, then the idea of remote backup service might be a differentiater. But you're actually late in the game for that idea. So, you need some reason why you're going to be better or different. You also are missing an important cost edge. A number of the players in the field already are big. They have server farms that have other purposes, and remote backup is a way of using surplus capacity designed for peak and excess demand. In other words, the added cost of server space for remote backup is virtually nothing. You can't beat that or their scale.

You should start out by investigating the competition. Research them in detail. Ask yourself what's wrong with them (what could be done better or different). Ask yourself if it is possible to beat their price, or, alternatively, offer something better that is worth a little more. To enter an existing field of competition, you need a niche or a competitive edge. Think from the outside. If you were a potential customer, why would you want to use your business rather than one of the existing businesses.

Unless you really have money to throw away, I wouldn't venture into this unless and until you have done all your homework, including the creative thinking, the technical analysis, the comparison to existing competition, and the financial business plan. Put it all together in a formal proposal and ask a friend who already runs a business to review it. Or, even better, present it to someone who funds startups and ask if they would be willing to fund it.

Just my 2 cents from having been there, done that.