Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Effectively using AWS Reserved Instances

stripe.com

Post reply on HN
Print view View original

71–80 of 123 posts

Re: Effectively using AWS Reserved Instances

#71

Post by jedberg »

hueving wrote: post #64
eldavido wrote: post #52

Earlier quoted context omitted.

I think this is a really illustrative example of how business strategy can influence decision-making. "Paying AWS's profit margin" assumes you can get access to their cost structure. You can't. In order to get AWS's cost structure, you need to (a) be buying servers by the truckload to get volume discounts, (b) have a scaled labor force for physically moving, racking, and installing all of this that's insured, directe…

Rediculous false choice. You make a huge assumption that aws charges as much as it would cost you to build your own without bulk discounts, etc. They charge way more than that. I've worked on 40-rack build outs using supermicro without any special pricing that beat the cost of AWS for an equivalent number of reserved instances.

Did you include your own salary and the salary of the person who did the POs and the person who racked them and networked then and configured the networks?

What about the salary of the person who maintains all of that? The cost of spare parts? The cost of downtime when hardware breaks?

Re: Effectively using AWS Reserved Instances

#72

Post by sonnyblarney »

taormina wrote: post #69
sonnyblarney wrote: post #51

Earlier quoted context omitted.

I don't see how arbitrary AWS instances are in any way going to go 'obsolete'. They have been around for a decade and are becoming more and more normative. Second, the underlying financial principle is that with visibility comes lower volatility comes lower cost - that's some very basic financial logic that's at play here. Yes, of course the contract implies a degree of vendor lock-in, but this is inherent in the nat…

No, the GP is referring to things like c3 vs c5 families. When you're locked into the 3 year old hardware, you have to pay the cost of upgrading eventually.

First - that 'lock in' is very evident by the nature of the contract.

Second - because newer things are available, it does not mean that others become 'obsolete' in three years by any means.

The vast majority of business do not need to have access to the latest, specific version of what are essentially commodity bits of hardware and software.

It's not really a risk for most businesses, and if it is, then obviously they can pay a higher price for the volatility inherent in switching at any time if they so chose.

Re: Effectively using AWS Reserved Instances

#73

Post by sonnyblarney »

boulos wrote: post #59
sonnyblarney wrote: post #44

Earlier quoted context omitted.

It's not so much sane, as possibly a little more friendly to those who can't - or don't want to do capacity planning. Visibility in terms of outcomes means savings, or rather, variability means cost. Ultimately, you're going to bear the cost if you cannot provide visibility because Google is not likely ever going to do it as well as you can for your own business. Ultimately, if you knew exactly what you needed over t…

Disclosure: I work for Google Cloud. > I think it would be nice to have Google's offer, but then also a longer term 'lock in' low price option as well, as frankly, this fits a lot of businesses. We hear you. That's why we offer Committed Use Discounts [1]. Are you saying that a 3-year commitment to a specific price (or lower, as we do price cuts) is insufficient though? (I want to understand) [1] https://cloud.google…

I'm only making a very general reference to the fact that long-term visibility and predictability entails lower cost and therefore lower price, and that business owners are likely more empowered to determine that outlook than the cloud provider, either AWS or Google. Ergo - some kind of customer oriented long term lock-in would likely, in the long run, produce the cheapest prices in the system. That's all.

Re: Effectively using AWS Reserved Instances

#74

Post by mmt »

Johnny555 wrote: post #55
jjeaff wrote: post #30

Earlier quoted context omitted.

Why would you compare AWS vs managing your own data center? You could also compare AWS vs building your own silicon. I think it would be better to compare AWS vs renting dedicated servers from a large provider? I think you will find that the scales tip heavily in favor of renting bare metal as far as price is concerned.

Why would you compare AWS vs managing your own data center? Because we were already managing our own data center. I think it would be better to compare AWS vs renting dedicated servers from a large provider? I think you will find that the scales tip heavily in favor of renting bare metal as far as price is concerned. We offloaded a lot of work to Amazon that we were doing ourselves -- database hosting, storage system…

> storage system management

I was a bit incredulous that any truly all-inclusive analysis could ever show AWS being cheaper, but this phrasing made me realize that it could have been the one (remarkably common) case where it usually does: enterprise hardware.

That world is easily more expensive than AWS, especially considering that hardware maintenance contracts are a thing (and a shockingly expensive one, to those of us accustomed to the commodity hardware world).

> Also, we liked the ability to have a failover region essentially for free - we only pay for enough servers to replicate the key data we need for failover, and keep the rest of the infrastructure powered off.

That's a useful advantage, though there's a pitfall in that there's no powering off EBS volumes.

Re: Effectively using AWS Reserved Instances

#75

Post by njovin »

msravi wrote: post #60

My experience with AWS reserved instances has not been very good previously. 1. Once you buy a reserved instance, you're locked in to that type and price for the duration, even though newer types at lower prices may get introduced (as they almost definitely would over 1-3 yrs). 2. If you're from outside the US, you might not be able to resell your reserved instance. So you're stuck with an old instance type at an inf…

I’ve gotten proactive emails from our account manager when they release new/cheaper instances and they offer us the option to transition and get a credit for our existing RIs.

We aren’t a huge account (less than 30k/month) so I thought this was a nice gesture on Amazon’s part.

Re: Effectively using AWS Reserved Instances

#76

Post by mmt »

jjeaff wrote: post #31
eldavido wrote: post #25

Earlier quoted context omitted.

If you're truly spending "a few million dollars"/year, then MAYBE. 200k/month would get you there. My personal breakpoints are 50k/mo PaaS->EC2/IaaS and then something like 100k or 150k start thinking about physical. Maybe. By the time you hire everyone, get all the planning right, etc. you might be there need-wise, but it's not a sure thing.

Why not just rent dedicated bare metal from a large provider?

I'd expect that to have nearly all the disadvantages of cloud (other than virtualization).

One is still locked into the provider's pricing structure.

One is still locked into the provider's ISP choices.

One is still locked into the provider's internal network architecture choices.

One is still locked into the provider's limited hardware choices.

The last one is the biggest one, if only because there are so many opportunities in so many components to maximize performance and minimize cost, with a little forethought and customization, all while staying well within the commodity market. The "one size fits all" model does many people a disservice.

This is especially true even at fairly low scale, where the issue of obtaining certain components in large volumes doesn't come into play.

I understand why some people insist it's just too hard to deal with hardware, but I find it disingenuous to advocate that opnion without at least a more comprehensive understanding of what modern, commodity hardware is capable of. I think most of the claims of difficulty are coming from those without this knowledge and have primarily software backgrounds (and are relying on hearsay or other second-hand experience).

Re: Effectively using AWS Reserved Instances

#77

Post by not_kurt_godel »

hueving wrote: post #58
eldavido wrote: post #20

Earlier quoted context omitted.

This is terrible advice for all but the largest of organizations. Running your own hardware is AWFUL. Get ready to dedicate an entire team to network engineering, fixing broken hard disks, patching operating systems, screwing around with RAID controllers, upgrading switches, planning power and cooling, and endless vendor negotiation -- with ISPs, hardware manufacturers, datacenter operators, etc. Oh, and did I mentio…

>This is terrible advice for all but the largest of organizations. Don't start a conversation with an opening generalization like that if you want something constructive. Especially when the rest of your post is clearly based on the single anecdote of your experience. >Running your own hardware is AWFUL. Maybe for you. Not for any sysadmin with even just a couple of years of experience. >patching operating systems We…

> Don't start a conversation with an opening generalization like that if you want something constructive. Especially when the rest of your post is clearly based on the single anecdote of your experience.

You are doing as much, if not more generalization by way of the assumptions you're making.

> >Running your own hardware is AWFUL. > Maybe for you. Not for any sysadmin with even just a couple of years of experience.

Sysadmins aren't real estate attorneys or facilities managers or security personnel. A lot of them aren't even tech ops who physically manage the DC hardware and are oncall 24/7 to fix problems on-site if needed. You need all of those things to run your own DC, and potentially a lot more if you're physically building the center itself (architects, contractors, civil engineers, etc.). And if you're serious about latency, availability, and durability, you're going to need those things in multiples for however many datacenters are needed to meet your targets. How many organizations have the millions of dollars of capex needed to get that off the ground and keep it all running?

> Netflix doesn't run 1/3 of the Internet traffic off of AWS, only a tiny subset because of the aforementioned shitty economics.

To how many organizations do the economics of serving 1/3 of Internet traffic apply? 2? How is that a counterexample to his point about datacenters only making sense for the very largest?

Even if you sidestep all those costs by renting instead of building and even if we take for granted that your "shitty economics" are still shitty down numerous orders of magnitude from Netflix-size, you're still burning money making your devs design and operate your system twice - once for the DC, once for AWS, and however much work it is to glue the 2 together. The end result may be cheaper infrastructure-wise but it will also be unavoidably less reliable and more complex purely by virtue of having more than twice as many moving parts.

Let's say implementing things this way takes a single dev time-and-a-half compared to just doing it on one or the other. Let's say (very conservatively) you pay your dev $100k a year + $50k benefits. You're now $75k in the hole from the get-go. That's enough to pay for roughly 83 m5.large EC2 instances on-demand (no RI) for a year. Your company has to be very large for the marginal savings of using a DC to outweigh that kind of deficit.

Re: Effectively using AWS Reserved Instances

#78

Post by sudhirj »

outworlder wrote: post #45
yani wrote: post #8

Earlier quoted context omitted.

Using GCC still feels like comparing an early android to a modern iOS. I guess they have to sell it cheap when the features and quality is not there.

Whatever GCP has is rock-solid and often superior to AWS. For example: when AWS encounters non-catastrophic issues with their hypervisor, you are on the hook for moving the instances away (meaning stop-start, or termination and relaunch for instance store). Depending on the instance type, this can cause service disruption. GCP will transparently migrate the VM while it is running for you. You never see it, you custom…

The AWS equivalent of Anycast / closest region is to route all traffic through Cloudfront. That way users enter the AWS fiber within 50ms (sometimes 5ms) and have SSL terminated there as well. Only works for HTTP(S) traffic, though, not general networking.

Re: Effectively using AWS Reserved Instances

#79

Post by mmt »

jedberg wrote: post #71
hueving wrote: post #64

Earlier quoted context omitted.

Rediculous false choice. You make a huge assumption that aws charges as much as it would cost you to build your own without bulk discounts, etc. They charge way more than that. I've worked on 40-rack build outs using supermicro without any special pricing that beat the cost of AWS for an equivalent number of reserved instances.

Did you include your own salary and the salary of the person who did the POs and the person who racked them and networked then and configured the networks? What about the salary of the person who maintains all of that? The cost of spare parts? The cost of downtime when hardware breaks?

Not the parent, but, I, too, have done the cost analysis, at much smaller scale.

And, yes, all-inclusive, AWS is 1.5x-10x more expensive than commodity hardware, depending on how poorly optimized AWS's hardware choices were for the particular workload and the commercial datacenter market at the time.

> What about the salary of the person

In general, I've found the need for the quantity of "person", bizarrely, exaggerated.

> The cost of spare parts?

Included, and it's low. Is this another aspect that's exaggerated?

> The cost of downtime when hardware breaks?

This is identical to the cost of downtime when AWS's hardware breaks.

Re: Effectively using AWS Reserved Instances

#80

Post by rohan404 »

One of the major issues we've seen with our customers is that many of them (especially startups and SMBs/SMEs) don't have the ability to dedicate a team to just managing their RI capacity. We've also seen enterprise customers optimizing up to 70% of their EC2 usage, but many of them have trouble ensuring a level of utilization due to rapidly changing infrastructure. I'd definitely argue that GCP has a better model for some use cases as it requires less active effort for optimizing billing, however if you manage your RIs on AWS effectively you can often get a better price. Looks like Azure has also gone down the same route as AWS, which is quite an interesting move on their part.

Disclosure: I head engineering/devOps at Engineer.ai - one of our products Cloudops.ai allows our customers to save up to 15% of their AWS bill without making RI purchases, as well as get discounted prices and additional flexibility (custom lock-in periods) for RIs they do wish to purchase. Feel free to reach out for information - my email address is in my about section.

Post reply on HN