for $10 a month for us low resource users. I agree that IO with 80 VMs will be high but for some of us it's bearable.

It's not going to happen. See this thread.

_________________
/ Peter

I've seen this come up a few times, the objections to this always seem to be centered around lowering the entry level price point. I can understand there are business reasons for this - Linode's business plan is presumably modeled on generating a certain revenue per user to ensure resources are available for support etc etc. In this case, introducing a lower price point is risky - you potentially get lots of new users, paying less money and all demanding support and quality suffers for all of us.

So - would Linode be interested in offering a lower spec/price linode as a second node for existing customers only? That would increase revenue per user, potentially provide use for older hardware and make quite a few people happy I suspect (I've tried cheaper VPS hosts, they ain't worth it).

It could also be good for Linode - if the new nodes were understood to be for beta-testing, you'd give people cheap access to beta environments (rather than risking their production ones) running new Xen stacks/kernels etc and you'll get a lot more feedback on them.

If you'd order one of these make yourself heard here - if we can convince Caker it would be profitable they might just do it . I'd take one right away.

While I don't see this happening, I do see some benefits and possible ways to make it work. As stated in the thread linked by pclissold, these types of servers could easily be used as small testing servers for use prior to moving code into the production 360+ server. They would also be good for use as small Nagios or other monitoring servers. They could also be good for very small, very low traffic web servers.

I also agreed with someone else that these could be sold as addons to existing servers, thus requiring you to already have a larger, more expensive Linode. That would help cover the costs of support. Instead of having 80 new customers to support for less, you would have 0 new customers with more income.

The last comment that I read in the previous thread was that somebody would cancel if Linode started overselling. That isn't the issue. There would simply be more virtual hosts per server, each with lower specs. That isn't overselling because they wouldn't be selling more servers with the same specs.

There are so many possibilities if only something were available. The specs don't have to be exactly half. Perhaps, slightly lower disk space or less bandwidth. I, personally, would use one for Nagios and as a central configuration repository for all of my other servers. Very little disk space and bandwidth needed.

Really people, "no" means "no". I don't see caker asking for ideas here - if you don't have at least $20 to spend, linode doesn't want your money.

Are you so sure? I'm sure somebody along the line said "We aren't going to have a plan under $200." If customers make suggestions and a company doesn't eventually listen to those suggestions then they aren't going to remain in business.

I'm not spending $20 a month on a Linode. I am spending more than that. That doesn't mean it wouldn't be nice to have a smaller test server.

Why bother with smaller?
imho - what Linode offers is worth 20 bucks a month even if it were a smaller server. heck, i learn enough good stuff from the forums, irc, and library that if i had 20 buck and didn't need a Linode i would still be proud to contribute.

For development / unit testing, wouldn't you be using a local virtual machine and for pre-production testing wouldn't you typically need more than the resources that a "180" linode would have? Right now today if you have say a Linode 720 for production and want to test in a pre-production environment before deploying you can;

1) Add another Linode 720
2) Copy images
3) Do your testing
4) Remove the previously added Lindoe 720
5) Receive credit on the remaining portion

That's roughly $1.29 per day for testing, which means you would have to do roughly 93 days of testing per year to justify the cost of the $10 Linode plan for a year.

Maybe I'm just jaded, but when I hear about ultra low Linode plans I start thinking IRC shell proxies and trouble users that can no longer get a shell account on a provider specializing in IRC proxies.

Not necessarily. I travel so much that it is easier for me to use a hosted virtual system for development. Not only do I not have a bunch of spare systems sitting around to install on, I also do not have the bandwidth for quick installs. Right now I am creating a set of install and configuration scripts and the Linodes are great for this. Sometimes I also use these systems to write various PERL and shell scripts Neither of which needs a lot of resources.

Right now I am paying for multiple 360s. I have a live server and two test servers, costing me a total of $60 per month. Each of these servers are used on a daily basis. Now, if I had one live server ($20) and one or two dev servers, then I would only be paying $30 to $40 per month. That saves me money and could save Linode money by allowing them to better allocate resources instead of having to add additional physical hosts to support 360s or 720s, etc.

I think these mini hosts could be limited to one or two per regular Linode just as IP addresses are. That way there aren't a bunch of people buying mini hosts It's just a thought though.

Why stop at 180? I want a Linode 36 for $2/mo. It'll have 400 servers all vying for the same hard drives. But I bet I could run ipchains in a 36MB environment and setup a nice little firewall. Maybe I'd also like it to have negative resources (no public IP brings the price down to $1/mo).

You laugh, but many people could do without a public IP address. For example, a back-end database server only really needs a private IP address. Being able to access it from another box or LISH might be acceptable; having machines in your Linode cluster that don't face the internet could be useful.

Hi,

An alternative to providing full mini-Linodes would be the ability to slice and scale the XXX in LinodeXXX. Examples: a couple of 300s and a 120 from a Linode720, or an 80 and a 280 from a 360.

The number of "slices" per Linode could be limited according to plan size: a 360 would be allowed 2 slices, a 540 could have 3, ... . CPU resources would be limited in the usual way so no harm will be done to co-hosted Linoders.

Revenue-wise Linode could profit from "slicing and scaling" by making it a premium feature and fill the gaps between the current price points: a sliceable 360 for $24.95, a sliceable 720 for $49.95, and so on.

I'd love this kind of setup - and it would make me stay with Linode forever

Cliff

Right now, CPU is not limited. If the current setup is used, anyone who sliced their node in two would have nearly twice the CPU and those who don't, because they'd have a minimum of 2/41 instead of 1/40 of the CPU.

GWAAAA Why is this still being discussed? It's not going to happen, please see this thread in which caker replied:



Not only are there resource issues to consider, but pricing issues.

[WARNING: Speculation and ranting ahead]

Processing credit cards over the internet costs more than physically swiping the cards. Keeping cards on file and processing them at a later time costs yet even more. The more risk of fraud the higher the cost.

Now, Linode may be using a 3rd party company that keeps everything on file and does the billing and who has an API so you can go through the Linode Manager to maintain your CC information and that may reduce their costs. I don't know for sure what Linode does. But below is some information for those of you who don't process credit cards and don't know about this to consider:

For every credit card transaction there is a fee for just swiping the card, plus a transaction fee that's a percentage of the total sale. Yes the retailer is paying the credit card processor on top of you paying the retailer sales tax (if you're in a place that charges sales tax).

I wouldn't be surprised if Linode is paying $0.25 (for "swiping" the card) + 3% of every transaction, possibly more depending on which card. Visa and MC typically cost the same. Discover is usually a little more and AMEX is usually the most expensive (by far). Though in my experience AMEX has a lower swipe fee and a higher percentage and depending on your processor, AMEX may be cheaper to process for smaller charges (less than ~$10).

Just using those numbers and their $19.95/mo price for the 360. They're paying $0.44 per every 360 customer per month. That's 2.2% of the total. Lower that price point to $9.95 and they're paying $0.35 per transaction, that's 3.5% of the total.

While that may not sound like a lot to you guys, it adds up big time. If they fall below the $19.95 price point, their profit starts to get eaten into even more than it already is with credit card transaction fees. This is why you see a lot of your tiny ma-n-pa shops require a minimum of $5 purchase to use a credit card, or "cash discounts".

Those places you go where there is a tip line, easily pay double the amount of fees than what the retailer who doesn't have a tip line pays.

I'm very surprised and happy with the level of service at Linode for their price points. It's not going to be beat by anyone out there. Find a cheaper (less expensive) provider and you're going to have much lower-quality of service. Heck, I'd be surprised if most can even provide the level of service Linode does at Linode's price points or even higher. Linode is one class act for sure.

If you want cheaper service, go to a cheaper provider and get what you pay for

Oh and if you're one of those people who use your debit/credit card for a $1.50 transaction at a small ma-n-pa, independent type shop, STOP IT. You're KILLING them. Depending on the product and their margins, you may have just cost them money.

If the shop you're at is a place paying for one of those check verification systems (every major grocery store for example) where they scan your check and debit it from your account almost instantly, it's not a lot different than processing credit/debit cards. Every check your write costs them money. Just use your debit card, the ONLY thing you're accomplishing by using a check is slowing down the people behind you and not doing ANYONE a favor, including you as you can't float the check anymore.

And what's the cost to process cash you ask? ZERO.

In a case where a LinodeXXX is sliced precisely in two, each slice's CPU minimum would be 0.5/40, not 1/41. ie. slices mustn't impact generally allocated resources. If I pay for 1/40 CPU minimum - that's all I get, even with slices.

At a general level, what I'm asking for is that just as we are already able to slice and dice our disk quota within each LinodeXXX, we should be able to do the same within the limits of our purchased CPU minimum, RAM and IO resources.

Slicers would suffer some drawbacks.

If you wanted to tweak a sliced Linode360 from "280/80" to "240/120", both slices would need a reboot. But if you left your "280" slice alone and then stopped or deleted your "80", your "280" would likely be left running with the resources of a "280".

The only other fairness issue I can see is the boot order of Linodes during a host restart. That could be resolved by having a "slice boot order priority" flag option in the slices configuration profile. Only the one slice with the "boot priority 1" flag will get booted along with the non-sliced Linodes. Once these have finished, the non-priority slice boot sequence can start.

Cliff