Please create an account to participate in the Slashdot moderation system

 



Forgot your password?
typodupeerror
IT

Ask Slashdot: What Would You Do With Half a Rack of Server Space? 208

Posted by timothy
from the give-it-a-piece-of-my-mind dept.
New submitter Christian Gainsbrugh (3766717) writes I work at a company that is currently transitioning all our servers into the cloud. In the interim we have half a rack of server space in a great datacenter that will soon be sitting completely idle for the next few months until our lease runs out. Right now the space is occupied by around 8 HP g series servers, a watchguard xtm firewall, Cisco switch and some various other equipment. All in all there are probably around 20 or so physical XEON processors, and probably close to 10 tb of storage among all the machines. We have a dedicated 10 mbs connection that is burstable to 100mbs.

I'm curious what Slashdot readers would do if they were in a similar situation. Is there anything productive that could be done with these resources? Obviously something revenue generating is great, but even if there is something novel that could be done with these servers we would be interested in putting them to good use.
This discussion has been archived. No new comments can be posted.

Ask Slashdot: What Would You Do With Half a Rack of Server Space?

Comments Filter:
  • Crypto! (Score:4, Insightful)

    by chucklebutte (921447) on Sunday July 27, 2014 @11:42AM (#47543711) Homepage

    Mine the shit out of any crypto that tickles your fancy!

    • install the rosetta @ home boinc project and predict and desing protein strucures.

      • silly. in current weather, i'd put a couple of pillows in there, take a Kindle and wait for the summer to pass me by.

    • I know how you can generate revenue! Sell it all to me for $100.
  • We need help in every form we can get.

    http://archiveteam.org/index.p... [archiveteam.org]

  • Keep It Ready (Score:5, Insightful)

    by binarylarry (1338699) on Sunday July 27, 2014 @11:45AM (#47543735)

    Keep everything ready, so you can switch back when the cloud services fail and/or your management team changes.

    • Re:Keep It Ready (Score:4, Insightful)

      by rogueippacket (1977626) on Sunday July 27, 2014 @12:13PM (#47543937)
      Pretty much the only sensible answer in this discussion so far - and based upon the number of people (trolls? shills?) saying that the gear should be used for mining crypto-currency, I could probably make a small fortune as a security consultant looking for abusive sysadmins wasting company assets for dubious gains.
      Let's not forget, your employer is moving to the cloud either because they do not see value in what you provide, or they want you focusing on more strategic initiatives. You should probably spend some time cooking up something amazing in the old environment or, worst case scenario, using it as an opportunity to brush up on your skills and certifications.
      • Re:Keep It Ready (Score:4, Insightful)

        by sjames (1099) on Sunday July 27, 2014 @12:38PM (#47544079) Homepage

        Or their management is from the cult of MBA and fears actually owning anything, or they just saw an ad for the cloud and got sparkly eyes and said "ooooooooh, shiny!".

        Meanwhile, a good admin will normally be just a bit bored because everything is running smoothly. It doesn't hurt if they have a zero priority thing to fiddle with as long as they continue working on the real mission.

        • Or their management is from the cult of MBA and fears actually owning anything, or they just saw an ad for the cloud and got sparkly eyes and said "ooooooooh, shiny!".

          I don't know of many small-to-medium sized businesses who migrate to the cloud because it's shiny. They all do it because they either read somewhere or were told by someone (most likely a salesperson) that it would save them money. Contrast that with the sysadmin constantly reminding them of the need for more hardware, more licenses, more overtime, etc.
          Anyone who tells you an IaaS migration is about something other than cost is probably trying to sell you IaaS. Fear of running your own infrastructure is j

          • by sjames (1099) on Sunday July 27, 2014 @02:29PM (#47544737) Homepage

            They move for the imagined savings. The shiny is what keeps them from examining the 'savings' as closely as they should. It frequently turns out to be much smaller once you see past the dazzle.

            It has it's place, it's just not all it's cracked up to be.

          • by 1s44c (552956)

            Few people use IaaS because they fear owning servers. Some people really like or need the flexibility. Some people have done the calculations and it does save them money. Some people don't want the hassle of looking after servers and want to concentrate on their core business.

            You try and build a reliable and fully redundant data center for less than amazon charges. It's not so easy.

            • by sjames (1099)

              I have been maintaining infrastructure for over a decade. I know what's involved.

              When you mention flexibility, you're getting somewhere. If I needed a temporary capacity bump, the cloud would make a great deal of sense. It's not a bad DR plan either. But for the everyday capacity (the base load if you will), ownership is cheaper and offers better control.

        • by 1s44c (552956)

          You really should reserve judgment until you know what his company does. For a lot of companies owning physical servers doesn't make sense.

        • If you EVER had to do destkop support 1/3 of your calls our HELP MY PST IS CORRUPT I MUST HAVE IT ALL BACK!

          It is great when the average person receives over +110 emails a day with a 100 meg quota is thrilling! People at work lose them all the time when their .pst hits 18 gigs and go all the way to SVP of IT to demand that billly gates fix it because they need every email for the past 10 years. ... ok rant off.

          But with the cloud quotas and .pst files are a thing of the past. At least I would want to outsourc

          • by sjames (1099)

            Nah, they'll just demand that you do something anyway. They'll go ballistic when the cloud service has a glitch and be absolutely certain you could fix it if you just tried hard enough..

      • "Pretty much the only sensible answer in this discussion so far - and based upon the number of people (trolls? shills?) saying that the gear should be used for mining crypto-currency, I could probably make a small fortune as a security consultant looking for abusive sysadmins wasting company assets for dubious gains. Let's not forget, your employer is moving to the cloud either because they do not see value in what you provide, or they want you focusing on more strategic initiatives."

        So those are the only t

        • Re:Keep It Ready (Score:5, Insightful)

          by rogueippacket (1977626) on Sunday July 27, 2014 @01:15PM (#47544305)
          Actually, no, there's only one reason any company moves to the cloud - because they think it will save them money. In-house disaster recovery is expensive. Employees are expensive. Refreshing hardware, licenses, and support agreements cost a lot of capital. The allure of trading all of that away for a fixed monthly cost is too strong to resist for most decision-makers.
          I don't want to sound overly bleak here, but anyone asking the Slashdot crowd for ideas on how to generate revenue for their employer using commodity hardware is probably so far removed the actual business that their days are numbered. Your Infrastructure was outsourced to an IaaS provider because they don't want to pay for the iron. Next, it's PaaS - your hypervisors, databases, and operating systems, and you with it.
          If you want some real advice, use it as a DR site (as GP stated) and make sure the business understands the risks associated with shutting it down, ensuring your ass is covered by having the CFO and/or CIO issue a statement to that effect (they will pin it on you when the cloud goes down regardless, because if you really read those IaaS contracts, the provider cannot be held liable). Then, walk away from it. Divorce yourself from the infrastructure discussions as much as you can, get involved with bigger and better initiatives so that once the salesmen show up with their PaaS offering, you're too well engrained in the big picture that they can't live without you.
          • Ether they are right or they are wrong (cloud services saving them money).

            For some companies the cloud will save them money and increase availability. Don't work for those companies as IT (janitor), work for them as IT (software developer).

            Outsourcing development is something else again. But until they succeed in outsourcing backups etc you can safely assume that development is a non-starter.

            In general outsourcers won't save you money if your IT department is competently run (outsourcers need to make

            • In general outsourcers won't save you money if your IT department is competently run (outsourcers need to make a profit), but for the 90% of other cases your job is in danger.

              What if your needs are small enough that having whole person doing it is 90% more people than you need? An outsourcing company could aggregate several businesses in the same situation, resulting in more efficient use of the staff and equipment.

          • by Sarten-X (1102295)

            I'm going to go out on a limb and say you're not a business manager.

            There are several reasons to switch to cloud services. Reducing current costs is one, but there are others.

            A business may be facing a market change. The IT needs may grow or shrink rapidly, depending on external factors. Rather than hiring extra personnel and planning servers for needs that might arise, and adding training to the burden of the existing admins, it may just be simpler to migrate to a cloud provider while needs are worked out,

    • Re:Keep It Ready (Score:5, Insightful)

      by multimediavt (965608) on Sunday July 27, 2014 @12:21PM (#47543983)

      Keep everything ready, so you can switch back when the cloud services fail and/or your management team changes.

      That was going to be my suggestion as well. I would not "get rid of it" or "donate it", Hell, I wouldn't let the lease expire either! I would keep that half-rack-o-stuff around for at least the next two years to see how well the "Cloud" does for you with the provider of choice. Plus, it never hurts to have a set of backup servers around that you control (that mirrors the data in the cloud, at least!). I have absolutely no faith in third-parties controlling my data and critical services. I might take advantage of some services but I would NEVER, EVER put my data under someone else's control ... did I say EVER? It's just a really bad idea and experience will teach you why. Good luck!

      • by jythie (914043)
        *nod* something we did in the past was when moving stuff to "cloud" infrastructure we mirrored all the input going to the cloud version back to the physical rigs. Not only did it give us a hot failover, but it also provided a good production-like environment for doing performance testing and wargames with realistic loads and data. If nothing else, having a production scale staging area to test things out in was a godsend since some stuff only starts going wrong when you have a lot of big systems talking t
    • by SuperBanana (662181) on Sunday July 27, 2014 @12:43PM (#47544113)

      Keep everything ready, so you can switch back when the cloud services fail and/or your management team changes.

      Did you miss the part about them trying to cut opex? *siiiiiigh*

      Even that aside...Maybe the latter, but not the former. One of the most common mistakes of failover environments is using the "old stuff" for failover/backup.

      That works great, until you exceed the computing/storage capacity/bandwidth of the original hardware.

      Let's say in a year traffic is up 30%. Something goes wrong, big time, with Teh Cloudz. You've done a good job of keeping the old hardware current and replicated. You 'flip the switch'...and the old environment promptly chokes...oops.

    • Re:Keep It Ready (Score:4, Insightful)

      by TechyImmigrant (175943) on Sunday July 27, 2014 @12:43PM (#47544121) Journal

      I'd be setting it up to keep a running backup of the data in the cloud, with the aforementioned 'keeping it ready' to serve from that data when the cloud gives way to sunshine.

    • by 1s44c (552956)

      Nah. Get rid of the kit and use management direction changes as an excuse to buy better kit.

    • by mlts (1038732)

      I might add on to that. Keep it ready, and if it does get pushed to the cloud, keep the half-rack as a "disastrous recovery" [sic] site.

      At the minimum, one can buy a small tape library (a single drive HP one that is 2-3U can store 300 TB, all encrypted, using LTO-6 tapes.) Add to this a 1U machine via a SAS card, and you now have archiving capability. A HP or Dell drive array is also 2-3Us, so add that onto the machine via your protocol of choice (SAS RAID 6), and now you have a place to stash critical d

    • by shri (17709)

      I seriously recommend this blog from Rackspace [rackspace.com] to those who are so caught up in cloudy-cluster-off-premises-corporate talk.

      This rising complexity and cost on the multi-tenant cloud is hitting customers in four main ways:
      - They spend more on engineering time and talent to architect for failure on the multi-tenant cloud, which is complex and hard.
      - They also spend more on engineering to deal with inconsistent performance, which is even harder.
      - They spend more on infrastructure, because over-provisioning is o

    • Keep everything ready, so you can switch back when the cloud services fail and/or your management team changes.

      Indeed. The cloud fad is already starting to pop as executives find out "Holy fuck, you mean when something goes wrong there's no amount of screaming I can do to make them prioritize our service?" and other things that weren't in the brochure. "You mean we're on a shared infrastructure so when one of the other tenants gets DDOSed we're down too? "

      Or (my favorite) "You mean to actually have high availability we have to spend almost double the quoted price to run identical machines in another geographic-zo

  • of course (Score:5, Funny)

    by Sergio CastiƱeyras (3767091) on Sunday July 27, 2014 @11:45AM (#47543737)
    porn, every flavor
  • Build a darknet. Maybe a TOR router? Donate CPU time to charity or some great number crunching project.

  • by Mikkeles (698461) on Sunday July 27, 2014 @11:54AM (#47543805)

    Make a mini-grow-op.They'll never flag the extra power used for lamps.

  • by Ken_g6 (775014)

    If you think good will for your company would go further than a few cryptocoins, you could do World Community Grid. [worldcommunitygrid.org]

    • by skids (119237)

      This suggestion would probably be the least work to set up and then tear down. Assuming the existing hardware is running a supported platform, it's just packages and a small amount of configuration and can run in an unprivileged account. When you get towards the end of the unplug date, start disabling new jobs from tasks with long-running jobs so you don't leave too many unfinished ones. And yes the WCG does have tasks that need storage, not just CPU.

  • Seriously? (Score:2, Insightful)

    by nospam007 (722110) *

    Ebay! An then you have the space for a pool-room, a porn-station, a man-cave or another dozen things with a dash in it.

  • do it

    • Yeah, 10 Meg isn't tremendous, so a Tor exit is probably as good as you can get. It's too small for a mirror host or a torrent seeder.

      I'm assuming you're unwilling to incur 95th percentile charges on your burstable. Tor allows easy bandwidth limiting right in the .conf.

      Still, that's only one machine - 10 meg is easy to saturate.

  • by NemoinSpace (1118137) on Sunday July 27, 2014 @12:02PM (#47543873) Homepage Journal
    The case has already been made against the assemblage of substandard HP garbage occupying half a freakin rack! The person that was blamed for this probably doesn't even work at this company anymore. The best thing you can do is contact the sales guy at the data center and form an unnofficial alliance and work out some preliminary arrangement for a kickback when you reopen your account after the cloud plan goes up in smoke. Most importantly, tell no one about this. You have obviously stumbled into another exploit of the BOFH. Stay out of the office till this blows over, you don't seem to be the intended target, and are messing with forces you don't understand.
  • According to legend at one Fortune 500 company I worked at in Silicon Valley, a data center administrator ran an Internet Service Provider (ISP) when dial-up was still king by using the spare server and bandwidth capacity. This gig went on for a number of years until someone in the corporate office noticed that the data center was far more active than it should have been and ordered an audit. The administrator skipped town and retired to Mexico as a millionaire before the audit got completed. Not wishing to
    • In the late 90s, Intel found one of the nets largest/most active porn sites hiding in the network.

  • mbs/Mbs (Score:5, Funny)

    by starless (60879) on Sunday July 27, 2014 @12:28PM (#47544023)

    I doubt you can do much with a 10 milli-bit per second connection...

    (Sorry, but I'm a scientist, units are important to me...)

    • As a scientist, you definitely want to be precise with your fractional bits.
  • Learn hadoop (Score:2, Interesting)

    by Anonymous Coward

    You have the makings of a mini-cluster there. Take the opportunity to learn to install/maintain/query hadoop.

  • Ignore them and get on with life?
  • I'd install Eucalyptus and develop an application. Then when my lease ran out I'd redeploy it on AWS.

    • Show me a 21U fish tank. Seriously. That would be awesome.

      • by i.r.id10t (595143)

        Tall and skinny... tricky to do.

        With appropriate lighting, etc. you could maybe duplicate part of a kelp forrest/bed w/ one or to plants going all the way up and then the wee beasties like sea horses, etc.

  • If it is in San Francisco area, take out the servers, furnish the rack, and rent it out.

  • If this were for my company, I'd want to do two things with the hardware. First, use it to back up the cloud environment. Maybe not the applications, but definitely the data. Disaster recovery is always paramount in the corporate world.

    Second, I'd want the hardware used to try out some new software, techniques, file systems, media servers, etc. It's never too late to learn new skills, and what better to learn on than servers you don't mind wiping if they get messed up. Using them to mine bitcoins is far les
  • by jmd (14060)

    Because it is not just a job.. it is an adventure!

  • by Giant Electronic Bra (1229876) on Sunday July 27, 2014 @01:32PM (#47544403)

    If its in say LD4 in London, or NY7 in New Jersey then I'd make a crapton of money leasing it out or selling VMs to brokerages. If its in ho-hum Dallas Rackspace somewhere or whatever then its not that interesting. Still, its a lot of iron to be idle in a big DS for that long. You could run a pretty serious web site on that sort of infrastructure. Maybe find some startup and leverage it, give them a leg up in return for some cheap equity. If it goes bust its no worse than leaving the rack idle and if it takes off you make some bucks.

  • I don't know if those machines are capable of competing for bitcoins, but it might be fun to try. Years ago I used to configure underused computers to do distributed computing stuff like SETI@home [wikipedia.org]. Now there are programs like Folding@home [wikipedia.org] where you can donate resources to medical research.
  • Beer! (Score:4, Funny)

    by ctrl-alt-canc (977108) on Sunday July 27, 2014 @02:00PM (#47544559)
    Remove the unused CPUs and disks, fill the free space with beer bottles, and take advantage of the cooling capabilities of the rack!
  • alarm spider web thingies to protect what's left. They seem to attract really cool burglars like Catherine Zeta Jones, Antonio Banderas and that guy from Ocean's Twelve.
  • If you want more servers like that, Wierd Stuff Warehouse [weirdstuff.com] in Sunnyvale, CA, has the same HP series G Xeon servers for $189. (2 3GHz quad-core Xeons per server! Hard drives are extra, but cheap.) Wierd Stuff has huge supplies of previous-generation data center equipment.

    It's amazing how cheap computer hardware is now.

  • Since you're asking this question, I'll assume that you have the freedom to do whatever you want with them. We'll assume they're your servers, personally. In that case, keeping them at the ready in case your cloud solution turns into a hurricane is a great idea that was mentioned previously.

    Otherwise, a couple of things come to mind:
    1) Start a web hosting company, using Linux and cPanel
    2) Start a Private VPN service
    3) Beowulf Cluster! (this is slashdot, after all...) or the modern version: OpenStack
    4) Profi

  • by mwvdlee (775178)

    Not a network expert, but wouldn't some sort of internet proxy with caching be a simple way to help?

    Other than that, the sad truth is that obsolete hardware is usually most productive when it's not using up valuable energy.

  • Run BOINC [wikipedia.org]. Discover pulsars, gravity waves, prime numbers, cure cancer. Isn't that enough?
  • Obviously.

  • Rent out the processing/storage as a cloud service!

  • I'd forget about it, let the lease run out.

    First, I'd be afraid that anything I tried to use it for would become 'mission critical' once it existed, then I'm the one 6 months from now saying "Yeah... we can't get rid of that as we planned in the budget..."

    Second, if you ever need something like that again, would you rather lease all new shiny stuff, or mess with rebuilding what you have as leftovers into something usable?

    Its legacy stuff. get rid of it.

    If you must, do something like run the Great Internet

  • and quit worring about wasting time on some goofy me me me project that's only temporary

  • Since you don't have a specific use for these servers, it's best to find someone who does. This could be a godsend for another small company that will be able to start it's services immediately rather than waiting for presumably more expensive new servers to arrive.

  • You should use them to host various flavours of your resume, because once everything is settled in to the cloud, your employment will be the next thing to expire.

  • You could consider donating server space and bandwidth to Openstreetmap [openstreetmap.org] projects. There's a wiki [openstreetmap.org] for OSM but it's quite confusing at times. Then there's Nominatim [openstreetmap.org], the name search; it requires lots of computer resources [openstreetmap.org]. Open source routing from OSM data can be done with OSRM [project-osrm.org], which is quite fast.

    Perhaps extracts of OSM data for downloading would be nice, eg. just roads, waters; see what's already available [openstreetmap.org].

  • You could use https://secure.slicify.com/ [slicify.com] to sell some of that processing power for real $.

Loose bits sink chips.

Working...