High-Memory Instances and $5 Linodes(blog.linode.com) |
High-Memory Instances and $5 Linodes(blog.linode.com) |
I use it to host an 8tb drive that I then attach to a dedicated server I rent for $20 a month.
$30 a month and I have a server that I use for Plex (16 cores, 3.2ghz, 32 gigs of memory, and 10tb of total storage).
Thanks for the recommendation, slot hosting is a good solution if you need to get a disk online with a VPS attached to it.
We've had a lot of people asking how you did this so just to explain:
Slot hosting $120/year ($10/month) - that gives you one 3.5" or 2.5" slot to send your disk in. You can aggregate up multiple slots and aggregate the VPS resources into one large VM. This is great for things like ZFS and software RAID. The VM is KVM based. https://www.delimiter.com/slot-hosting/
The server is this: Dual E5420, 16GB RAM, 1TB or 2x500GB HDD $200/year ($16.66/month). https://cc.delimiter.com/?cmd=cart&action=add&id=1718
Slot hosting uses dedicated (non-contended) CPU/RAM/Disk resources so you can run Plex directly on it if you have enough slots aggregated to give you the RAM.
That sounds amazing. Where can I find this?
Personally, I settled with colocation. I pay $60/mo + $2k one-off for the initial hardware + say $150/5y/4TB HDD, which, for 80TB of storage over 5y comes out to a total of ~$88/mo, or $0.001/GBmo. Even if I was to store 3 copies of everything instead of doing erasure coding, my costs are roughly half of what Backblaze B2 charges. And my disks are fully online block storage, not necessarily object storage.
From the blog post you just commented on ;-)
One of our easiest pricing wins is against say the i2.* series ("Were you trying to buy more flash or more cores?") and now against their p2.* GPU series. It's pretty liberating to actually mix and match.
Disclosure: I work on Google Cloud.
But after using them for 12-18 months, and losing several days of data due to the 2015 DDoS, and reading about more and more security issues, I switched back to DO and haven't looked back. The performance differences aren't noticeable to me, and I'd rather have my hosting through a company with a better security record than Linode.
But that doesn't work with Linode plans because they keep increasing the resources every now and then. They doubled the RAM on all the low-end plans a couple of years ago. Before that, they doubled the storage.
Another option would be to use random IDs that point to a specific bundle of resources. But then the plans change, and the previous bundle is no longer available for purchase.
It's hard to use consistent identifiers when the resources they point to change so often. Linode customers absolutely love those free upgrades.
I bet there is a piece of code in Linode codebase that has the plan resources mapped in a table, with a list of ID attached to the various sizes. The genius developer decided that it makes sense to have the "growing number ID" match the "growth in resources" assigned to the plan. In other words, the bigger ID should always mean MORE resources.
This leads them to alter IDs if they have to introduce plan-sizes that are small.
Imagine if AWS did this: you would see riots in the streets!
Other than the Christmas 2015 DDOS, I have never had any significant issues with Linode. Especially in the Dallas DC.
That being said - their policy for handling DMCA notices is abysmal. They will network filter out your Linode until you respond to the notice.
Hopefully they at least give you some warning before a filter is placed.
- A prohibition against the propagation of data, images or sounds that may constitute defamation, an insult, denigration, or an infringement of privacy, image rights, good morals, or public order
- Users are required to use decent and respectful language
- Users are reminded that they must update software without excessive delay when a security failing is noted by the user or the software publisher or Scaleway
This reads like: If you might offend anyone by the standards of the French, or might use strong language, or don't run the latest version of everything, this isn't the service for you.
On the positive side, they lack the litany of restricted services that most hosting companies I've seen provide, and I don't see any prohibitions against using your paid for resources to their maximum, so that's good.
US customers should note that all of their prices are in Euro, so -- for exchange rate shenanigans and currency exchange fees on most credit cards. There's no selector to show prices in USD.
I haven't heard much about these terms being held against users, so I'm not sure how much action there would really be behind this type of stuff. With that said, it is definitely more comfortable to be on a host that doesn't have terms that are this strict.
With the help of Ksplice the Ubuntu Server droplet has achieved 555 and 401 days uptime without downtime (could hang on a bit longer but later decided to reboot once every 3 months to address security concerns).
DO support has been responsive and friendly, DO keeps (slowly) delivering new features such as private network, Load Balacer etc. For existing $5/m DO users, I don't think it's worth the hassle to migrate to Linode (or Amazon Lightsail), the performance difference will be unnoticeable for most people's use cases (personal web hosting, strongSwan based IPsec VPN, etc.).
A good reference: https://joshtronic.com/2016/12/01/ten-dollar-showdown-linode...
Will provide the feedback to DO and see if they can match the Linode offer (I am sure they will do something).
Since my posts are straight benchmarks, I don't touch on the support side of things typically (mostly because I only have had support interaction with the companies I use).
That said, I've had exceptional support from both Linode and DigitalOcean over the years. Always responsive and friendly.
Interestingly enough though, yesterday my Linode went down due to a power outage in the Atlanta data center. I'm a huge fan of Linode, but something like a power outage at this stage in the game seems a bit like amateur hour.
I enjoyed that link to the $10 showdown. Linode's VPSs usually preform quite well. It did take Linode a while to get into the SSD game, but when they did, they got the IO right.
Redis appreciates it.
Well, I'm really questioning what you've used then. It's atrocious. It's right up there with GoDaddy in terms of dashboards that are needlessly obnoxious.
Digital Ocean's may be simple, but there's nothing wrong with that. It works. It's clear what it can and can't do. It's not cluttered up with confusion.
For example, on Linode you cannot delete a Linode instance anywhere but the main view. You must go back to the main listing, carefully look for the one you want to remove, then click the remove link and double-check you clicked the right link. If you have a lot of instances and you cycle them over frequently enough this is a real hassle.
Likewise, there's many occasions where you get kicked back to the index page for no reason. There's just so many unresolved little things that, over time, grate on you considerably. It works but it could be considerably better. It has not evolved much since launch, that's very concerning.
Digital Ocean's interface, to use one example, has evolved considerably. Amazon's AWS dashboard may be a monstrosity but it's also becoming better and better organized over time. Linode needs to remember that their dashboard is important and invest in it.
Maybe all you ever eat is porridge and you're okay with that. Fine. Other people demand some real food now and then.
Having to go in and change my password is hardly an exceptional event on today's internet.
When you search for "linode hack," Google suggests "linode hacked again." They were hacked in 2012. They were hacked in 2013. They were hacked in 2014. They were hacked in 2016. They will be hacked in 2017.
Maybe it's possible to scale with just RAM! Fire up a uWSGI + Flask and up the RAM as traffic increases.
If the server is up, even if you use 0% CPU, you're getting charged.
Per-use is a lot cheaper for people that spin up servers to meet demand and kill them once things slow down.
"If My Linode is Powered Off, Will I Be Billed? If your Linode is powered off, but is still added as a service on your account, you will still be billed for it. This is because Linode maintains your saved data and reserves your ability to use other resources like RAM, transfer, etc. even when your Linode is powered off. You will be billed for any other active Linode service, such as Longview Pro or an extra IP, as well."
[0] https://www.linode.com/docs/platform/billing-and-payments
They introduced per-hour billing recently, that's the only fundamental change to their billing, and that was something you had to go out of your way to opt into if you wanted it for an existing account.
If you provisioned a $20/mo. machine then yes you will pay $240 per year for it. This is how math works.
until very recently, they didn't allow using a custom kernel (except via kexec hackaround) and were quite slow updating their kernel for security patches. they repeatedly gave random dates for implementation, then repeatedly pushed them back, then eventually just ignored users on this issue for years.
their images were also poorly sanitized, leading to the well-known problem of SSH host key duplication, which was the case for years.
Source? Just because they haven't announced any successful security breaches, doesn't mean they haven't been seriously targeted.
In fact, considering the amount of times my random dedicated server instance (not hosted at DO) gets hit with random attacks, I'm sure a large provider like DO has had numerous serious, targeted attacks against their network/servers/control panel/etc.
http://venturebeat.com/2013/12/30/iaas-provider-digitalocean...
As the official authority on serious attackers I can confirm that his is in fact not true.
This was heartbreaking because Linode's value (performance, reliability, support, price) is almost impossible to match if your requirements fit their VM configurations.
But in the end, we did not want to risk another security or DDoS fiasco. We estimated that the risk was high that they would be targeted again, we could not believe their promises to get better at face value considering previous transparency issues, and we did not want to tell our customers that a company that had experienced security issues for the past three years had suffered from another attack[1].
They seem to have invested quite a lot in their networking infrastructure (kudos!), but I believe they still use their old coldfusion applications.
[1] Non-technical customers often ask us why we are not hosting on Amazon because they heard it's where serious companies host their servers (!). We used to explain why Linode was a more cost-effective choice, but Linode was not a nice name to google in early 2016.
For now, yeah. But the new Manager mentioned in the blog post is just a client for our new API, which is written in Python.[1]
You can even check out the new Manager now if you want: https://github.com/linode/manager
It's still in alpha (and separate from our regular service), but we're hoping to have it out in beta soon!
[1] https://engineering.linode.com/2016/04/12/Announcing-APIv4.h...
I don't particularly care if they keep around a 512MB/$2.50 instance at that point.
60$/month for 16GB, 1CPU, 20GB disk for VPS... in comparison a root server with real remote console access, 16GB, 4 core AMD, 4TB disks, unlimited traffic costs me 30EUR/month (30$/month).
Or for 60EUR/month (60$) I could get a root server with 64GB Ram, 2x500GB SSD, i7-6700, 30TB traffic.
OVH et al can get you servers quickly if they have them "in stock", but supply fluctuates wildly, especially on the cheaper end.
All this does depend on the VM host actually implementing these things properly, but they are out there.
DO is slicker, but it's not 8k/year slicker. I can do better things with that money.
https://www.theregister.co.uk/2017/02/06/cisco_intel_decline...
I bought some new IPs 4 hours ago for my VPS. I've had an automated response to say they've taken payment. Do I have the IPs? Not according to the control panel. I imagine I'll get them at some point in the next 24 hours. shrug If not, I'll put in a support request to ask where they are and maybe wait another 2 or 3 days for a response.
When I signed up a few months back, I purchased my first VPS on a Saturday. At no point did they tell me that my order needed to be manually checked. I eventually got my VPS on Monday morning.
Does anyone even work there on the weekend?
Amateurs
That was it. They also own Kimsufi. I had one account there too. Cancelled both of them and moved to VPS providers and it's been 3 years w/o complaints (my usages were/are very small time though).
When I checked around on hosting forums, looking for ways to contact them, everyone used to laugh as in "Support and OVH? Lol".
I have one for a few months and according to htop, the VM is up and running for over 90 days.
If you can't get together the people for a full cabinet (which I think is definitely worth it, the more you buy at once, the cheaper it gets), there are a number of providers that offer single server colocation. The only one I've really heard much about though is Joe's Datacenter, which is in Kansas City, Missouri.
Unfortunately there isn't really a "Linode" or "DO" in the colocation space anymore. There used to be https://prgmr.com, which offered transparent pricing based on power and bandwidth usage but they've stopped that I believe. The few global providers are generally geared towards multinational corporations or at least businesses more than small groups or hobbyists. You're really best off with the smaller regional providers near wherever you are.
Every major urban hub has a multitude of providers. Some offer expensive, but reliable service, others cheap but you get what you pay for, and there's a mix in between. Talk to people and ask for tours. Every town is completely different in terms of scene.
It wouldn't have cost them anything to tell me up front that IPs aren't automatically provisioned and required manual intervention and it would take ~24 hours before I get them either. Luckily for them, this time I didn't log a support ticket, because I just assumed that they would have crap processes and that's why it was taking so long. I mean, why did it take so long? Where they manually checking my order again, even though I used the same payment details? Does somebody need to manually pick some IP addresses from a spreadsheet? Why is this process not entirely automatic?
This is the sort of thing that should be automated for low cost hosting systems, precisely to prevent having to provide unnecessary support.
I'd rather something like ImpactVPS or Linode, as with either I know things will be decently managed, unlike OVH. Don't get me wrong, those VPSes make great dev boxes or Tor relays, they just have no performance or reliability guarantees AT ALL! Do not risk your business on them, spend a few minutes to find either a smaller provider with good management (generally at the same price point or lower), or pay up for Linode or AWS.
Link: http://www.webhostingtalk.com/showthread.php?t=1470503
True, and in fact the only basis for my statement is that they have historically taken security so not-seriously that it would be surprising if they were in fact able to withstand advanced attacks, given that even the most secure organizations are often unable to do so. (see: every talk at Black Hat)
> In fact, considering the amount of times my random dedicated server instance (not hosted at DO) gets hit with random attacks, I'm sure a large provider like DO has had numerous serious, targeted attacks against their network/servers/control panel/etc.
this statement is just as baseless as mine. perhaps even moreso, since the two numbers seem to have nothing to do with each other. one could just as well say "my server gets lots of bogus SSH attempts, so banks get robbed a lot".
You can take a snapshot of a VM, you can move it to a different host, you can spin up a new instance to test a major upgrade, you can clone a cluster for a few hours of testing, you can scale up and down...
Unfortunately, this is fairly standard practice in the industry. Companies want to make sure the vulnerability is closed, positively identify what was compromised, who was affected, what legal liability exists, and so forth.
Weeks is, frankly, pretty quick to go through that process.
> There's not much more to say -- they are dead.
Huh. Funny, I'm still hosting things there; their prices are competitive, there are no rumors of acquisition or shutdown... Seems quite alive to me.
> They will be hacked in 2017.
And, as I stated originally, I have no reason to think they will be unique in this.
https://news.ycombinator.com/item?id=10845278
There's a clear, undeniable pattern of incompetence here.
I also suggest reading this glassdoor review: https://imgur.com/sJd56AT
And this thread: https://news.ycombinator.com/item?id=11136743
This is so far beyond malicious and incompetent it should be illegal. What you really mean is "Are we obligated to report this to our customers, or can we cover this up and get away with not letting anyone know we've (maybe/probably) been hacked?".
Customers' entire businesses are on the line. As in, a company can literally go bankrupt and/or be forced to shut down if the hack affects them. The only acceptable resolution is to warn customers within ONE HOUR of knowing that their company MAY be at risk of a hack. One day is already too late. A week later means that no pre-emptive mitigation was even possible, and it's simply too late to even try and protect oneself.
IT IS STRAIGHT UP NOT ACCEPTABLE, TO NOT IMMEDIATELY INFORM A CLIENT OF A __POTENTIAL__ THREAT TO THEIR BUSINESS. ___POTENTIAL___, NOT ___CONFIRMED___.
Companies like Linode are so busy trying to cover their own PR asses, that they don't understand just HOW CRITICAL it is that their clients be instantly informed of any potential threat. They think their business's reputation is important, without having a single clue that their entire business's success relies on their clients' businesses being safe. Informing all their customers that there is a 0.00001% chance that their account has been compromised FAR OUTWEIGHS the eventuality that even a single account was in fact compromised.
They just don't get it. They are prioritizing their own business's PR over their clients' businesses' well-being. And so a single hack reported weeks after the fact, without any early warning having been raised, completely destroys all credibility. A hosting provider should be put out of business after a single such failure to immediately warn clients of even a remote possibility of a problem.
tldr; Providers like Linode who prioritize confirming their liability, before so much as even considering issuing a warning to their clients that they may have been compromised, should not be allowed to do business. It really is as simple as that, and frankly anyone who continues to host with a provider that failed to raise any warnings until weeks after a potential hack deserves whatever business-destroying event happens to them next time. You cannot trust a company once they've purposely postponed releasing crucial details of an incident. Quite literally: in the future, when you find out you've been hacked on Linode a month after the fact... what the fuck did you expect? Precedent indicated this was the likely outcome... you got exactly what you stayed signed on to experience!
I realize that I am doing that thing where you try to rationalize your preconceived notions, but here are some high-level observations:
* Looks like there are normally setup fees associated with provisioning new servers, although they are suspended at the moment.
* Support seems next to non-existant, which I suppose is not surprising considering they're a low-cost provider.
* SSD-based servers appear to be frequently out-of-stock.
* If you need a KVM attached to your server, it is $30 for 24 hours or $200+ for a week.
* 250Mbps bandwidth (presumably in and out) cf. Linode which is 40G in and 1-10G out.
All that being said, my interest is piqued. I could see this being a good fit if 1. you have a more-or-less dedicated sysadmin, 2. need a lot of storage and/or memory and have solid sizing requirements ahead of time, and 3. cloud (VPS or otherwise) isn't an option due to cost or other facts. It could be great for running your own VM or container cluster. Thanks again for sharing!
https://www.hetzner.de/us/hosting/produkte_rootserver/ex51ss...
Dual E5420, 16GB, 1TB HDD (or 2 x 500GB): https://cc.delimiter.com/?cmd=cart&action=add&id=1718
Slot hosting: https://cc.delimiter.com/cart/slot-hosting/
Its not cost effective to have customers who want to load up on 'warez' cancelling their service, shipping back their disk to get it back the next week empty for another round.
We're positioning this as a long-term storage product where customers don't want to be paying the hosting company each month for disks.
Seriously, people do not understand leasing and renting.
All their marketing during the change was how it was cheaper. When it clearly was over double the price. And your argument is "it is cheaper if you are not using". So good luck going to a hotel and not booking any room so it can be cheaper.
It's only cheaper for companies that have lots and lots of servers, but most of them aren't provisioned most of the time.
But if you want something hosted/online, then it gets harder.
Presumably you can select the relevant ID based on plan properties (cores, RAM, etc.) rather than hardcoding IDs. Though I completely agree that once a plan has a specific ID, as long as that plan is available the ID should be constant.
What if they upgrade the 1024 plan to 1536? (That has actually happened, by the way.) Now you're extracting the number and comparing it to a range?
> Give me the plan where the RAM is >= 1024 and < 2048.
What if they upgrade the 1536 plan to 2048?
> Nevermind, just give me the plan with two cores.
Okay, do you want the $20/mo plan or the $120/mo plan?
> Fine, sort the plans by cores and memory and give me the smallest plan.
Oops, they introduced a new plan with one core and less (i.e. not enough) storage.
I'm not saying it's impossible. Just impractical.
I currently use Linode, but if DigitalOcean offered something at that pricepoint they would be getting $2.50 more from me a month than they currently do.
The blog should be updated now, might need a refresh to see:
https://blog.linode.com/2017/02/14/high-memory-instances-and...
I use one of my DO instances as staging for file transfers. 2 days a month, I can barely get more than 30K/s from their sf data center to comcast. Other days, I get 5M/s. Their network is shitty.
And I don't think I'm being rate-limited; it's 1-2G transfer per day.
DigitalOcean might happily be able to push 10g to their next hop, but beyond that things are significantly more outside of their control. This is mistaken as 'throttling' a lot, but is actually just ISPs not investing in peering to handle their peak demand. Similar to the Netflix and Cogent issue that happened a couple of years back.
At times, I can get better speeds by tunneling or proxying using a DigitalOcean Droplet and downloading something from overseas than I can doing it directly. The path taken from my home internet to the DO Droplet uses different peering than my path to Europe, and the speeds are faster overall even though it creates more hops.
So it appears that 75Mbps was the initial speed but after some internal deliberation (possibly influenced by feedback), they raised it to 1000Mbps.
[0] https://blog.linode.com/2017/02/14/high-memory-instances-and...
I'm not sure about Linode
The API should throw an error, so I can adjust my scripts, instead of silently provisioning a server I never intended to provision.
There should be a deprecation process when a plan is discontinued, so this doesn't happen unexpectedly.
If AWS can't provision me a t1.micro, they don't just go and give me a m3.small instead.