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.
Small scale VPS Providers
I recently got a funding of $10k to startup a VPS hosting company. Im in the midst of getting dedi deals from multiple providers.
To the VPS providers out there, mind sharing how much of VMs do you host on a single E3-1240, 32GB RAM, 1G uplink KVM node?
Dont worry about having another summerhost, we will be paying all our servers on a Yearly basis.
Thanks!
Comments
You got funding but dont have idea how virtualization works?
Smell fishy
Yes we do know how virtualization works and infact we already have a lab running in the background. What i would like is to gather real world information of how many guests are hosts putting on a KVM node.
@auriga
Congrats on funding as that is not something easy to get for this industry unless you are a big guy.
I am guessing if you are going with an E3-1240 you would be doing openvz with that you would be looking between 20 to maybe 100 customers depending on the plans the customer pick and what resources they really use. As well as how much are you willing to over sell the server.
If you are looking to do KVM then starting out with something like Dual E5-2620v4 128 GB Ram 4x 1 TB SSD would get you going then you can always add in more ram and more drives depending on the server you get.
Are you planning on leasing the servers or you want to buy the server and just colo? Either way need to pick a great up line provider that would be willing to help out since you are just starting out.
It's different for every provider. It all depends on the deals you release and the customers you attract.
Your best bet is to start slow and do small scale tests. Offer a free beta trial for 1 month while you gather data.
Shoehosting is on the air!
to the providers around here. how much does a VPS company earn in profit/year? I am guessing there is no money in the LE market. Where do you get the profit? Mid-tier+ customers?
I don't think a company that's been in business for a long time would really give out their estimated profit.
Got a loan, this is going to be tough for you my friend.
We use similar servers for Xen VPS nodes and really the onlyvthing holding back hosting more users is disk IO. Chances are CPU/RAM wont be an issue.
If your users are buying 1GB RAM, then you can probably do 30 VMs without any issue as long as your disks are fast enough. You will want SSDs, but if you need more space, you can consider HDD in RAID 10.
Your node only needs a modest amount in overhead itself.
16
@auriga As you have fund, invest on buying few server and IP space
You must have a plan of execution of each penny and decide when the investment can taken back
I had gone to develop without funds. I have started with $2 and now here with hand full.
LET is good for OpenVZ. Let me explain little(MOD remove if something offending the rules)
I have a ovz pack advertised on LET. Server spec not that fancy E3-1230(few L5520 also),32GB RAM,4x1TB HDD, with /26(from a leased /24) - barely comes around $90
Getting a /24 and assigning to server with separate subnet will cut the cost of running server. Dont dream you can offer good service with 256 VPS in a server.
ovz:
we have 40+(even more if we have resource )ovz per server. The issue arise when it hits the I/O wait as its not on hardware RAID with BBU.
On KVM :
LET pack : Above the RAM level for the LET pack(only linux). without that we wont get profit.
Other KVM: We offer windows on normal pack. So things gets trouble same on high I/O and core usage. Have much free RAM and disk. But CPU exhaust (I am clueless even after this much years in industry how others put 10+ windows VPS in a E3 and running smooth).
Industry is full of trademarked undocumented secrets, Experience few years to make a comfort earning. All the best
32GB ram will get you 128 VM's for sure
I'll raise you. 16GB for 256 VMs.
It depends on your model, but any 'well organised' provider will lose (a LOT) for the first few years - and that is with 'normal' margin clients, not Low End.
It took dediserve more than 3 years (and x,XXX,XXX's) to reach profitability.
Go ahead and get your kidney pre-removed to pay back the loan so your body gets used to having one
+1, If you plan to use KVM on this CPU then you won't get anywhere near what's some people say, well that is if you don't want your clients VPS to run like shit,
I gather you will run vCPU not dedicated as the obvious answer if your running Dedicated CPU is how many cores you have, as for vCPU we only allow 2vCPU per core which is again how i imagine that @exception0x876 got the figure 16 from, correct me if I'm wrong
That still gives 128MB. If they're OVZ, they should be just fine in 96MB. That'll give you ~340.
You forgot 5:1 overselling on top of that. With some deduplication on ram you could even reach 10:1 if you dont like your customers too much.
If you force them to only use Debian Wheezy, you can probably fit 400 easily before the oversale..
64MB CrapNAT™s inbound ;-)
Don't ;-).
Well for kvm you just divide out the resources. But to be honest that is not a good setup for a KVM node that you actually plan to sell services from. Maybe for personal use with a hand full of VMs but if your going to be charging people money and don't want to loose that 10k you should 1. Get your own hardware. 2. Get the right hardware for the job.
You said a bunch of words.
You can use Squeeze and squeeze even more.
Bourgeoisie proletariat be sated!
You can be profitable in the first year as a small guy who does things right, but that needs a lot of experience, so, probably, wont work for you. I suppose a decent chance starts from the second year.
A fairly big on the small side provider, such as one with some 4 k customers and some 10 k VMs can hope to have a decent net income (after taxes, wages, traffic, power, lease costs and everything else) from it, like 4-5 grand a month at least, being conservative (i.e. keeping the network clean, not too many customers in at a time, long term relationships, etc.)
In 5 years, it can be done with 10k start-up capital.
The alternative is big expansion and race to the bottom on prices, sell after two years some 2k customers.
One can make money in the low end market but I doubt that that's feasible for a newcomer, particularly one with small pockets.
And you have problems to overcome. Things like disks, no matter whether hdd or ssd, which are cheap but I/O limited. Or bandwidth; Unless you stick to small boxes you'll either need multiple uplinks or offer lousy guaranteed bandwidth.
Doing it based on rented dedis will almost certainly kill you because all the positive factors (like cheap colo or 2nd hand hw in good condition, etc.) end up in your providers pockets while you pay a high price.
About the only chance I see for someone in your position and w/10k$ is if you had LOTS of know-how and experience and contacts - which you do not because you ask here - and/or a niche that most can't fit into, typically based on know-how, again.
It can be useful for a fast ramp-up at a cheaper cost and then later transitioning to owned hardware when you have higher revenue. Though it helps if you pay setup fees to lower the monthly cost.
Leased equipment is the better solution, IMO. It worked for us and I think it is cheaper than rented overall.
Considering this is LET, I'm surprised by the number of people here who think $10k isnt enough to get something off the ground.
Regardless of your budget, knowledge and 'know-how' are a must for success in the industry. This is pretty much true of all businesses... Experience is a plus but not strictly required for success.
$10k is not in itself a lot of money but it is also by no means chump change. More than the $10k, a start up needs dedication and time. If you want to start a successful hosting business, you need to know what you are doing and allocated sufficient time to implement your plans and support your clients.
Depending on your model, renting servers may be the way to go since it requires significantly less upfront capital. Using your own hardware and co-locating does save you money in the long term but you will need significantly more capital if you have even a modest customer growth rate.
These days, margins are relatively small, and people are selling cheap VPS for less than $5 per GB of RAM. Assuming 64GB RAM on a node, and assuming you compete at LET levels, you kind of cap the earning capacity at $300 USD /month and you'll probably need $1,500 for the hardware, and probably close to $200 for the colo space, IPs, and bandwidth. Less if you can provide your own IP space (but that also has upfront capital requirements) and for $250 /month you can probably start finding dedicated server providers offering such servers.
So renting servers and IP space allows start-ups to compete at the lower end price points with minimal upfront capital investment, with only marginal increase in operating costs. If the startup is successful, then you can start looking at getting your own IP space, getting IP transit, and using owned hardware (not necessarily in that order). To save money with using economies of scale requires a reasonable volume of business.
So @auriga, do not feel pressured to provide your own hardware. If you plan to sell cheapo servers with small margins, you will either need to stick to reselling, or come up with much larger amount to invest. If your model is higher margin, and you can recoup the cost of the hardware relatively quickly, then of course $10k may be enough.
There is no 1 formula for success.
Good luck OP!