This is more of a policy issue than a technical one, but it is unfortunate that included bandwidth stops scaling proportionately to Linode size/cost after the Linode 4096 level. The net result is that buying two LN 4096's is a much better deal than one 8192. The cost is the same, but you get far more bandwidth (and more IP addresses) with the dual 4096 configuration. So my organization has to rearchitect our system to work well with multiple 4GB Linodes even though we'd rather simply upgrade to a single 8GB or 12GB instance.

Here is the page which shows the discrepancy: https://manager.linode.com/signup/#plans

If Linode can afford to offer 3200 GB of bandwidth when we buy two Linode 4096's, why can't you afford that when we spend the same amount of money on a single 8192?

I'm a huge fan of Linode and I hope you will consider this request! A great start would be raising the transfer cap from 2,000 to 3,200 GB for plans 8G and higher.

Thanks!
Gordon

From the comments section of this announcement two years ago:



Personally, I think it's a bit of a silly restriction, because nothing stops a customer from buying multiple Linode 4096's and pooling all the bandwidth. But maybe they had a strange case of abuse in the past that made them think twice about it...

As caker's message said, there aren't too many people that have this problem right now. If I were you, I'd contact support and see if they can work with you on this or if that limit is set in stone. They might be willing to make an exception.

I'm rather confused - what's the disadvantage to Linode if you get a single 8192 instead of 2x4096 when looking at bandwidth?

There may be a point in future where I need to upgrade our mail server from a 4096, however it's slightly annoying that I'm going to get less value for money by upgrading.

4G gets 1.6T of transfer, 8G/12G/16G/20G get 2T of transfer.

If you're not bumping up against your transfer limit, it wouldn't matter to you.

Transfer is the one resource that has remained more or less the same over the last few years, despite multiple free upgrades on RAM and disk space.

I suspect that there are two reasons for this:

1. There's relatively little pressure from customers to increase the transfer quotas. The vast majority of Linoders probably use only a small portion of their transfer, whereas RAM is always scarce and numerous people have complained about disk space.

2. Linode doesn't oversell RAM and disk space, but they might oversell transfer. In fact, it would be stupid from a business point of view not to oversell it, because (a) the vast majority of customers only use a small portion of it, and (b) overselling transfer to a certain extent does not have any effects on network performance.

If overselling is the case, we can think of the smaller plans' generous quotas as a free bonus, and the larger plans' stricter quotas as reflecting something closer to the real cost. Or rather, real cost plus a comfortable profit margin.

If you ever actually reach the point where you bump up against your transfer quota, just add another 4096. It's always a good idea to have redundancy if you're operating at such a scale, anyway.

Another way of saying it is ram/disk is a fixed cost (capex) while transfer is a variable cost (opex).

Thanks, everyone, for all the replies. I was really hoping that Linode would step up to the plate and adjust the LN8192 bandwidth package to be equal to two 4096 nodes. But they haven't responded :/.

As it is, we're charged a $240 monthly bandwidth overage fee for every 1600GB of excess bandwidth used. Yet we can buy LN 4096 plans, including 1600GB of extra bandwidth, for just $160 per month. Even less if we sign up longer term.

To avoid the huge overage fee this month, my org had to buy 3 more Linode 4096's, which we probably won't even power on.

I'm very happy with the Linode service overall and want to stick with them, but It is a shame that the bandwidth pricing isn't more rational. I'd love to upgrade our main system to 8G or 12G instead of buying a cluster of 4G nodes that we can't even use with our current architecture. But then we would run out of bandwidth and suffer the monthly overage fees, which are considerably higher than even Amazon EC2 charges.

They didn't respond to a support ticket? If you haven't opened a ticket, I'd certainly do that (as suggested earlier in this thread) - the answer may still not be what you want, but I'm pretty sure you're going to get a response.

If you did open a ticket and they didn't respond (a) I'm really surprised and (b) I'd add another comment to the ticket. I don't think I've ever waited more than a few minutes at least for an initial response to any ticket, so not hearing anything seems like it has to be an oversight/mistake.

-- David

I second @db3l's suggestion that you should open a support ticket about your unique situation. This forum is not the official support channel.

Despite the name "Feature Request/Bug Report", Linode rarely responds to feature requests on this forum. I'm sure they monitor it pretty closely, but I suppose it's their policy not to intervene in discussions among customers unless absolutely necessary.

You can also prepay for bandwidth at $0.10/GB/mo from the Extras tab, which will get you (as an example) 2000 GB/mo for $200/mo. I believe it auto-renews, as well, but don't quote me on that.

_________________

/* TODO: need to add signature to posts */

Until you cancel it, yes, it automagically renews!

Many of you suggested that I open a support ticket, so I gave that a try. Linode did respond quickly, but they only offered to "pass this feature request along to management for their reference."

Hoopycat noted that you can prepay for 2000GB of bandwidth for $200/month. But that seems mispriced because you can buy 10 Linodes 512's for $199.50, and those includes the 2000GB of poolable bandwidth along with a whole bunch of storage, RAM, and CPU time. And if I prepay a year in advance, that cost drops to $179.55. Buying the bandwidth by itself costs $200/mo even if I prepay.

So for now I guess I'll have to keep buying extra Linodes for their bandwidth allocation. I'm currently only using one of my four 4096's. Still, I hope Linode will consider including more bandwidth in their larger plans rather than capping them all at 2000GB/month. Given that a Linode 16GB costs twice as much as an 8GB, it should include more bandwidth. But it doesn't.

Finally, I'd like to thank everyone who responded to this thread. Even though my problem isn't solved, I'm glad to have at least gotten it off my chest .

Look at it this way, few applications need more than 4GB of RAM in a server that can't be load balanced among multiple boxes, and in this case, you can get each of your linodes in a different datacenter and set up a high availability setup for better reliability

Can you imagine a Beowulf cluster of these?

_________________

/* TODO: need to add signature to posts */