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.
Comments
We still have Opteron HE with DDR2 in production, hundreds of them. For 90% of our users it's perfect. Cheap storage and bandwidth, with just enough CPU to get everything done.
We even offer than as dedi seedboxes upgradeable to regular dedis: https://pulsedmedia.com/managed-dediseedbox.php most of them are actually 48GB but since we cannot be arsed to upgrade old servers from 24GB to 48GB anymore we advertise all as 24GB. DDR2 is flaky AF, takes hours sometimes to just get a set of functioning 48GB.
Those pretty much are sold out as soon as we have some available on the lower capacities.
For those who need the CPU grunt we have the ZEN variants
Typically you do not calculate end cost like that, no matter how many times the hardware has been paid off already you STILL have to make a profit on the hardware, and get payment for it. Whatever your chosen ROI is, all you do is update to current market rates really for that hardware.
So something like HW Costs / Months to ROI + Electrical + Housing + Expected average BW per unit + Support + TRX Fees + Misc Admin costs == your minimum price for selling them. How precise you get this calculation determines your competitiveness more or less.
Yeah but you're selling a different product. Yes the base fundamentals are built on top of a dedicated server and you can make the argument that "well they can SSH in and use it however they wish", but the actual product is a managed dedicated seedbox. You have a value-add on top of old hardware to a client-base that typically don't care what the hardware is as long as it's on, has a fast connection to the internet, and can download/seed without "problems". You're also removing the management requirement to keeping things running, which again is a value-add. I've previously purchased seedboxes before not caring about the hardware spec but rather having someone else deal with the management side. I don't think that's a valid argument for this.
Yeah I never said you don't make a profit on the hardware. I've been generalizing costs to 4 major cost centers:
Power, Bandwidth, Space, and Hardware. Salary/Support is also a major cost factor but, for the sake of this theoretical argument, I'm factoring that into cost of operating the space (or cost of the bandwidth for a network engineer, etc.).
Since we're focusing on cost-based adjustments, as the customer you have to recognize where the best "fudge-factors" are, or what areas you can find room to ask for some relief. Hardware that's EOL usually means you have a bit more room to work with or at least have a discussion with the host you're willing to work with, because you're not going to be reducing power, bandwidth, or space cost on a rented server anytime soon.
There's more than one way to crack an egg. Each person operates their business using different parameters, but at the end of the day you have to make sure the fund balance works. Hardware costs is where you can still find reductions while making the fund balance argument work is all I'm saying. Whereas OP's situation they'll have more rigid hardware cost requirements that don't have the availability to be reduced. You want your host to make a profit so that they'll continue to stay in business and work with you (moving shit around is just annoying). But you also want to reduce your own costs. So you work something out.
Most users decide to choose the addon for regular dedi, it removes the server from our automation and turns into self managed regular server -- for extra money. That's like 2/3rds maybe 3/4ths of those who purchase it.
You can say that preinstalled seedbox software is the value addon tho.
sorry i must've misread then, i thought you meant hardware for free.
At super cheap servers it's the labor which often gets more expensive, all the QA, racking, cabling, furnishing, software etc. it takes surprising amounts of effort. well at least for a small operator like us, since nothing can be mass produced too low volume for that.
Yes, everyone has different cost basis, and different characteristics they need from servers, which changes the equation.
Well as you know the cost of QA, racking, cabling, etc. are all base costs that don't change based on server cost or hardware. 1 Server == 1 cost of QA, racking, cabling, etc. That's the cost for you to colocate your hardware with either remote-hand or someone on-site. So having a cheaper price tag means a higher percentage of the revenue goes towards paying those costs. I mean you already knew that but figured for others. I used to work in hosting (before LET had a provider tag), so I'm definitely aware of the costs that goes into operating these things.
That's an interesting statement then. Assuming the seedboxing communities, I'd assume most are willing to pay a little extra to get additional "privacy" and to retain as much control over their service as they could. Maybe that's what incentivizes them to take the preinstalled seedbox software (the value add) and then update whatever they need from then on. They just needed help getting it set up.
Still though, I don't think that service is really a good comparison as it's targeting a different market with different feature sets. But maybe that's just me being stuck in my own ways.
@HalfEatenPie going for the "Novelist" tag?
Don't get me started on Adaptec ...
Maybe I just got lucky but my experience with Dell was largely positive.
Uhm, there basically is no algorithm. It's ultra-simple slicing and mirroring. If you experienced Raid 10 cards being slower than mdadm then something was very wrong with either that card or your measuring.
No need to tell me anything bad about HP boxes. Whatever it is, I'll believe it ... (I avoid their hardware as best I can).
That may all be so but in this case we're talking about low-end storage dedis. And the typical LET user (incl. myself) approach to this is: how much storage cap, how much bandwidth, and how much monthly traffic for how many $$? And that's about it. @Calin's boxes can be attractive, IF their price tag is low and if they are in a halfway decent DC.
As for regaining the cost of hardware: given that both boxes are in the same DC and with the same connectivity, the difference boils down to an 26xxv2 box being purchased for about $1k less than a v4 box (with comparable fillings/features inside). Calculated over 3 years that's a diff/mo of roughly $25 - and that's a significant difference because it's one of the not so many fixed cost factors. Yes, I get you, that's a ridiculous amount between adult players - but not in the LE segment; there it may well be what makes your offer work or not. A box costing $25 less per month will sell, one that costs $25 more might not, not easily and quickly anyway. And for a cheap storage dedi even a crappy stone-age AMD 10 or 15 Watts processor (like the ones on PCengine boards) or C27xx processor would be good enough.
See: https://en.wikipedia.org/wiki/Non-standard_RAID_levels#Linux_MD_RAID_10
I cannot seem to find where the differences between HW and MD RAID10s, tho HW controllers may have catched up by now. Difference was how the RAID10 is laid out, MD providing higher throughput. Essentially it was 0+1 not 1+0
also even "ultra-simple" is an algorithm at the end of the day
The algo is how it's laid out on the drives.
lol, i hear you. I try to stay as far as possible from HP stuff ... except for once they have HW i want, they have some nice cheap 12G SAS HBAs which have been getting appraisals. Typical HP tho, have to know the abstract model number to order them. Once they arrive we shall see how well they function in reality.
Exactly this. Many of our seedboxes are being chosen by those numbers too.
Sometimes it gets even worse than that, we used to get flak on certain R forum for using shitty ol' CPUs, but no one was willing to double the price to get new modern ones.
Today's hardware example: 3x 12drive chassis' with v1/v2 CPUs and tons of RAM for ~200-250$ a pop, vs a EPYC 36drive monster at more like 5000$ a pop (probably more)? Let's say you get 18TB drives for just 300$ a piece, that will make 11 550$ vs 15 800$ - 17.82$ vs 24.38$ / TB purchase cost. 24mo ROI: 0.7425$ vs 1.015$. 36.70% price increase.
But relatively the EPYC will have more users affected by issues, because it's single machine. Also much less ram and bandwidth. So apart from CPU, users get less of everything else -- even IOPS if we tailor towards the low end segment on the drives too.
3x 12drive will consume about 750W while EPYC uses 400W is one of the few cost benefits.
You can get cost/TB only so low, that's almost fixed cost in the sense you cannot get any quantities of super discounted drives even refurbished, there's a floor on the price per TB set by new mfg storage. Demand is high and constant.
Just trying to give the full context. There's always someone who says "but X!" and "but Y!" which usually is avoided if you give full context.
There's where I think we have a bit of a concern. Of course if the pricing is attractive then people will come. It's the law of the market. But my argument is that there's no reasonable way (unless he's using second hand HDDs or decides to stop paying DC fees) to get those servers to a level where they're considered "affordable" and a "great deal" to the LET community in a reliable way. Now the definition of reliable is different from person to person, I mean some might be more accepting of the risk of data loss than others and are happy to pay less to take on that risk, but my argument is that factoring all these in, you can probably find, for the same "expected price" of the product that was proposed, a server with better specs and the same amount of hard drives.
Yes people's primary concern is storage volume + network bandwidth, but if you had two choices, one's the server configured the way OP suggested, and another that's the same but with better CPUs and newer RAM both for the same price, then which would you choose? I'm willing to bet that as long as supply is there on the newer hardware, people will choose that over the older hardware.
Now pricing strategy (or someone who just knows the rules of economics) would definitely price the newer hardware a bit higher than the older one, to prevent them from competing against each other (or for whatever strategy they're using). But that's assuming if it's just the same hosting provider. If two providers are competing then I think it's a solid strategy for the other hosting provider to target the same price point with sightly better hardware will take most of the new clients.
That's the benefit I'm suggesting people should be looking for and, therefore, stating that demand just won't be there to generate a return to pay off the $$$ plus profit.
But that's not the reality. Reality is this: All but the hardware costs are pretty much the same. Yes, I hear you, the older processor probably consumes more el. power. So what, clocks can be lowered ... et voila, you are within the power price bracket again.
So the real choice is this: get a more modern (and maybe more reliable) machine for more money -or- get an older machine for less money.
Pricing strategy sits on top of financial basics. OpEx being roughly the same, $1000 CapEx difference leads to a significant price difference for the customer, period (we're talking LE market).
Also keep in mind how many customers tick; they e.g. usually don't care about expected hw lifetime; after all they are buying a service and in time slices, so if one day say a memory stick fails it's not their problem. Again, we're talking LE market. Of course in other markets customers tick differently and may expect and be ready to pay a premium for a super-reliable service (and indirectly for super reliable hardware, n+1, etc.)
We do have some E5's in our RTO program now and are looking at adding more RTO servers in the very near future.
I actually saw that the other day! That's cool to hear!