Results 1 to 23 of 23
  1. #1
    Join Date
    Dec 2008
    Location
    Florida
    Posts
    1,052
    I've looked on the forums and Google, haven't been able to find much detailed information on this subject.
    Would some people please expand this topic just a bit?

  2. #2
    Join Date
    Dec 2004
    Location
    Chicago
    Posts
    1,000
    N+1 generally means that for as many pieces are needed for operation (UPS, generators, HVAC, etc.), there's one extra for redundancy. N+2 would be two extra.
    Some also throw around 2N+1, which would be one extra for every two that are needed. That's really about it.

  3. #3
    Join Date
    May 2005
    Location
    New York
    Posts
    737
    wouldnt that be double + 1. one extra for every two needed would be should be N+N/2 (also can be written as 3N/2)

  4. #4
    Join Date
    Aug 2000
    Location
    Sheffield, South Yorks
    Posts
    3,627
    Correct 2N+1 is actually double the amount needed, then an extra as well. i.e. If you need 3 generators to run the facility if grid power goes, then you'd actually have 7 generators: (2 * 3) + 1 = 7

  5. #5
    N+1 is like RAID5
    N+2 is like RAID6
    2N is like RAID10 and so on
    where N is a specific service, like AC, UPS, Generator - who knows what else...

  6. #6
    Join Date
    Dec 2004
    Location
    Chicago
    Posts
    1,000
    wouldnt that be double + 1. one extra for every two needed would be should be N+N/2 (also can be written as 3N/2)
    Yep, you're right. More coffee.

  7. #7
    Join Date
    Dec 2009
    Posts
    2,297
    Quote Originally Posted by qwidjib0 View Post
    wouldnt that be double + 1. one extra for every two needed would be should be N+N/2 (also can be written as 3N/2)
    Yep, you're right. More coffee.

    So wait is 3N/2 better than 2N+1?
    REDUNDANT.COMEquinix Data Centers Performance Optimized Network
    Managed & Unmanaged
    • Servers • Colocation • Cloud • VEEAM
    sales@redundant.com

  8. #8
    Join Date
    Mar 2011
    Posts
    122
    Quote Originally Posted by KDAWebServices View Post
    Correct 2N+1 is actually double the amount needed, then an extra as well. i.e. If you need 3 generators to run the facility if grid power goes, then you'd actually have 7 generators: (2 * 3) + 1 = 7
    Actually most datacentres use 2n+1 to say 2*n+1 so if you needed 3 generators to run the facility they would have 8 (2*4)

  9. #9
    Join Date
    Jan 2011
    Location
    UK
    Posts
    776
    You'll also see references to Uptime Institute Tier ratings, which are much more closely defined than any 'N+' specs. These are (roughly speaking):

    Tier 1 - no redundancy
    Tier 2 - some redundancy but still single points of failure
    Tier 3 - concurrently maintainable - ie you can power anything down for maintenance and there is enough redundancy to keep things running (colloquially, and not always accurately known as N+1).
    Tier 4 - double everything - sufficient redundancy is built in such that even after powering something down for maintenance, there is still redundancy in the system (colloquially 2N, actually at least 2(N+1)).

    Tier 1's are generally cheap and cheerful, basement type operations, Tier 2 covers more operators than you might imagine, or indeed that might admit to it! Personally I would say 'enterprise level' (if that means anything) starts at Tier 3. Tier 4 are normally used mainly by financials and governments as the cost is of course much higher than the lower Tiers.

    The tier definitions are widely abused by data centre operators but still provide a good indication of resilience in a facility.
    Advania Thor Data Centre Iceland - www.thordc.com
    High Spec Colocation and Dedicated Servers, powered by cheap, abundant and 100% renewable energy.
    Enterprise grade hosting, ISO27001 accredited for security, and all at fantastic pricing.

    brian.rae@thordc.com

  10. #10
    Join Date
    Dec 2009
    Posts
    2,297
    Quote Originally Posted by Brian_R View Post
    Tier 2 covers more operators than you might imagine, or indeed that might admit to it! Personally I would say 'enterprise level' (if that means anything) starts at Tier 3.
    I would agree to this. A lot of providers will claim Tier3 or say N+1 and really do not meet many of the requirements to achieve "concurrent maintainability"
    REDUNDANT.COMEquinix Data Centers Performance Optimized Network
    Managed & Unmanaged
    • Servers • Colocation • Cloud • VEEAM
    sales@redundant.com

  11. #11
    Join Date
    Mar 2010
    Posts
    65
    Quote Originally Posted by connected2 View Post
    Actually most datacentres use 2n+1 to say 2*n+1 so if you needed 3 generators to run the facility they would have 8 (2*4)
    missed maths classes?

  12. #12
    Join Date
    Mar 2011
    Posts
    122
    Quote Originally Posted by greecejoe View Post
    missed maths classes?
    urm would you like to expand?

  13. #13
    Quote Originally Posted by connected2 View Post
    urm would you like to expand?
    I'll do it for him.


    Quote Originally Posted by connected2 View Post
    Actually most datacentres use 2n+1 to say 2*n+1 so if you needed 3 generators to run the facility they would have 8 (2*4)

    you said 2N+1 right? And 3 generators to run the facility?

    So let's substitute the 3 generators for N, then it will be 2*3+1, you do multiplication or division first. so it would be 6+1 = 7 generators.

  14. #14
    Join Date
    Jun 2011
    Location
    Columbus, Ohio
    Posts
    32
    Note what Brian_R clarified above. It's not 2n+1, but rather 2(n+1). As Brian explained above, it's about remaining redundancy when one line (or, most always, a component in that end to end line such as utility or generator or transfer switch or UPS or PDU or strip or cord) fails. With n+1 (Tier 3), there's a redundant line that keeps power flowing...but now that 2nd path stands alone and any of its components represents a single point of failure (but overall, a good level of redundancy). With 2(n+1) (tier 4), each component in each line is also redundant so that if a component fails or is removed from service, there is still total redundancy. Generally speaking, Tier 4 is twice the system of Tier 3 (and Tier 3 is already a lot).

  15. #15
    Join Date
    Jun 2011
    Location
    Columbus, Ohio
    Posts
    32
    to clarify, utility and generator are alternatives of the same component (power source) in each line
    DataCenter.BZ
    Ohio's high-density, carrier-neutral colocation and virtual infrastructure provider

  16. #16
    Join Date
    Jan 2012
    Posts
    76
    Quote Originally Posted by Columbus DataCenter View Post
    to clarify, utility and generator are alternatives of the same component (power source) in each line
    Not quite sure what you mean by this, but technically TUI doesn't even look at / care about utility service or infrastructure. they simply assume it's not there, and just look at support infrastructure.

    this ultimately means you can have a Tier IV data center with only a single utility feed, as dumb as that is...

  17. #17
    Join Date
    Jan 2003
    Location
    Chicago, IL
    Posts
    6,957
    Quote Originally Posted by Columbus DataCenter View Post
    Note what Brian_R clarified above. It's not 2n+1, but rather 2(n+1). As Brian explained above, it's about remaining redundancy when one line (or, most always, a component in that end to end line such as utility or generator or transfer switch or UPS or PDU or strip or cord) fails. With n+1 (Tier 3), there's a redundant line that keeps power flowing...but now that 2nd path stands alone and any of its components represents a single point of failure (but overall, a good level of redundancy). With 2(n+1) (tier 4), each component in each line is also redundant so that if a component fails or is removed from service, there is still total redundancy. Generally speaking, Tier 4 is twice the system of Tier 3 (and Tier 3 is already a lot).
    But that isn't what was said, he said:

    Quote Originally Posted by connected2 View Post
    Actually most datacentres use 2n+1 to say 2*n+1 so if you needed 3 generators to run the facility they would have 8 (2*4)
    2*N+1 ≠ 2(N+1) What he said was the first, what he meant was the 2nd. I think you clarified it properly, but just pointing out what was actually said, probably just nit-picking at this point.

    To note, I do agree with the general knowledge, being N+1 does NOT mean you're Tier 3, in case, MOST N+1 configurations would be Tier 2, and I would imagine the vast majority of facilities that WHT customers are in are Tier 1 and Tier 2. Tier 3 would meet most enterprise standards and Tier 4 is really only used for or needed for true mission critical uses, like financial exchanges, major banks banks, etc. that can afford spending twice as much on a facility to save 5 minutes a year (on average) of downtime. It upsets me when people throw around the terms without caring what they mean, like MediaTemple claims to be in Tier 4 facilities, yet they're in DuPont Fabros Facilities, which are possibly Tier 3, and certainly not Tier 4, as they have many N+2 (or similar redundancy level) services. DFT doesn't claim they're Tier 4, yet their customer does, with nothing to back that up. To be Tier 4 you can't just have some, or even most systems entirely concurrently maintainable and able to sustain failures, EVERY system needs to be able to.

    Note: That is just one example, but there are MANY more.
    Karl Zimmerman - Founder & CEO of Steadfast
    VMware Virtual Data Center Platform

    karl @ steadfast.net - Sales/Support: 312-602-2689
    Cloud Hosting, Managed Dedicated Servers, Chicago Colocation, and New Jersey Colocation

  18. #18
    Join Date
    Aug 2004
    Posts
    93
    Quote Originally Posted by KarlZimmer View Post
    But that isn't what was said, he said:



    2*N+1 ≠ 2(N+1) What he said was the first, what he meant was the 2nd. I think you clarified it properly, but just pointing out what was actually said, probably just nit-picking at this point.

    To note, I do agree with the general knowledge, being N+1 does NOT mean you're Tier 3, in case, MOST N+1 configurations would be Tier 2, and I would imagine the vast majority of facilities that WHT customers are in are Tier 1 and Tier 2. Tier 3 would meet most enterprise standards and Tier 4 is really only used for or needed for true mission critical uses, like financial exchanges, major banks banks, etc. that can afford spending twice as much on a facility to save 5 minutes a year (on average) of downtime. It upsets me when people throw around the terms without caring what they mean, like MediaTemple claims to be in Tier 4 facilities, yet they're in DuPont Fabros Facilities, which are possibly Tier 3, and certainly not Tier 4, as they have many N+2 (or similar redundancy level) services. DFT doesn't claim they're Tier 4, yet their customer does, with nothing to back that up. To be Tier 4 you can't just have some, or even most systems entirely concurrently maintainable and able to sustain failures, EVERY system needs to be able to.

    Note: That is just one example, but there are MANY more.
    @KarlZimmer
    2(N+1) = 2N+2

    What tiers are your DCs?

  19. #19
    Join Date
    Oct 2011
    Posts
    1,321
    Quote Originally Posted by KarlZimmer View Post
    To be Tier 4 you can't just have some, or even most systems entirely concurrently maintainable and able to sustain failures, EVERY system needs to be able to.
    Sadly, this is very true.

    Too many hosts will say "Oh look at us, we're 2N+1" when their only talking about 1 service, but leading customers to believe that they are talking about all services.

    Your host can have 2N+1 power, but N+1 on data and only N cooling. This does not mean they are 2N+1!! Only their power is. The rest of the infrastructure still has gaps in redundance, that can bring your services offline.
    RDO Servers - Shared Hosting, Resellers, VPS, Dedicated Servers, & Secure Clusters
    Windows Hosting with Plesk Onyx - Linux Hosting with cPanel - LiteSpeed - CloudLinux - MariaDB
    Providing Excellent Hosting since 2009!

  20. #20
    Join Date
    Dec 2009
    Posts
    2,297
    Quote Originally Posted by RDO Servers View Post
    Sadly, this is very true.

    Too many hosts will say "Oh look at us, we're 2N+1" when their only talking about 1 service, but leading customers to believe that they are talking about all services.

    Your host can have 2N+1 power, but N+1 on data and only N cooling. This does not mean they are 2N+1!! Only their power is. The rest of the infrastructure still has gaps in redundance, that can bring your services offline.
    This is an old thread, but still very true. Lots of providers claim Tier 4, or some sort of high redundancy / concurrent maintainability and it is simply not true. Most of the time you do get what you pay for.
    REDUNDANT.COMEquinix Data Centers Performance Optimized Network
    Managed & Unmanaged
    • Servers • Colocation • Cloud • VEEAM
    sales@redundant.com

  21. #21
    Join Date
    Apr 2004
    Location
    Singapore
    Posts
    1,234
    Again this apply all the way up to the software stack.

    Hosting in a fully redundant tier4 (2n+1) data center will not achieve higher uptime if your software cannot run in 2N+1 mode.

    Many user deployed a single server in a tier4 data center and expect the server run 100% uptime, they forgot that the single server is the single point of failure.
    Alan Woo, alan [@] ne.com.sg
    = NewMedia Express Pte Ltd (AS38001)
    = IP Transit, Colocation & Dedicated Servers in Singapore | Hong Kong | Tokyo | Seoul | Jakarta |
    = Singapore Speedtest speedtest.sg

  22. #22
    Join Date
    Jan 2003
    Location
    Chicago, IL
    Posts
    6,957
    Quote Originally Posted by dr_lucas View Post
    @KarlZimmer
    2(N+1) = 2N+2

    What tiers are your DCs?
    Tier 2 and Tier 3 depending on the facility. The number of users needing to be in a 3rd party facility that actually need a Tier 4 facility is extremely small and a specific niche market.
    Karl Zimmerman - Founder & CEO of Steadfast
    VMware Virtual Data Center Platform

    karl @ steadfast.net - Sales/Support: 312-602-2689
    Cloud Hosting, Managed Dedicated Servers, Chicago Colocation, and New Jersey Colocation

  23. #23
    Join Date
    Jun 2011
    Location
    Portsmouth, UK
    Posts
    327
    Tier's and N's dont really interpolate.

    For example, a DC could be Tier II but still have 2N+1 UPS.
    But if those UPS's are fed from the same panel having that feed split before its presented to your rack only protects you from UPS malfunction

    Tier IV requires two independent feeds all the way, I.e 2N cables and switch boards.

    As others have said have Tier IV 2N+1 redundancy in theory takes you to a statistically safer place than Tier III with 2N+1 but Tier III is quite reliable, even Tier II requires something to fail (which can be mitigated against if correctly managed).

    Tier IV also doesn't guarantee any kind of connectivity quality, no point having 2N UPS and 1N connectivity.

    Tier IV also don't help if there is a fire and the fire brigade turn up and shut down the whole facility before they'll put the fire out.

    Finally, the Uptime institute award both design and operation separately, no point having 300 UPS's in no one knows how they work.

    If you want true protection you need to be in two different locations with two different providers fully mirrored in real time ideally a good couple of hundred miles apart.

    For >90% of applications Tier II is probably sufficient, Tier III is a good to have and Tier IV is an overkill.



    My advise would be to ask questions of your provider.
    ServerHouse | Est 2001 | 3x UK Data centres | Roof access, satcoms | High density | DR as standard
    http://www.serverhouse.co.uk

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •