Page 2 of 2 FirstFirst 12
Results 26 to 46 of 46
  1. #26
    Join Date
    Feb 2011
    Location
    Denver, CO
    Posts
    124
    Quote Originally Posted by iTom View Post
    There is no clear definition of what a cloud is - however that's basically it.

    Some add in features like an API, also fast upgrade/downgrades are also common (to make the VPS scalable)
    Yeah, that's the problem. I think it's sort of obvious though. It's called a cloud because you have a lot of hardware interconnected. So two web servers in a farm is not a cloud.
    Mike Kauspedas - GearHost
    PaaS Cloud for .NET and PHP Developers
    Publish using Git, Visual Studio 2013 or FTP. Signup - No credit card required
    Personal blog www.mikesaysmeh.com

  2. #27
    Quote Originally Posted by srfreeman View Post
    Cloud hosting is being touted as redundant, high availability hosting that alleviates problems from hardware failure. Conversely, none of the problems that users ask for solutions to here are caused by hardware failure. Most problems discussed here stem from programming issues exacerbated by the lack of knowledge on both the consumer and provider side. The discussion revolves around a solution without a problem and a problem that is never attacked at the root level.
    There are different kinds of users who will run into different types of problems. If you're competent and running a large organization's IT department, you aren't using CPannel, and making noob mistakes, but if your website goes down for any reason you are in deep ****, which is where the market for scalable and extremely reliable hosting platforms comes in. 100% uptime (no, it's not impossible, it's just very difficult) is something people will pay good money for.

  3. #28
    Join Date
    Nov 2009
    Posts
    544
    Quote Originally Posted by Question Everything View Post
    There are different kinds of users who will run into different types of problems. If you're competent and running a large organization's IT department, you aren't using CPannel, and making noob mistakes, but if your website goes down for any reason you are in deep ****, which is where the market for scalable and extremely reliable hosting platforms comes in. 100% uptime (no, it's not impossible, it's just very difficult) is something people will pay good money for.
    I agree that "scalable and extremely reliable hosting" is being sought. Most "cloud" hosting offerings today are simply servers built as vms that have cPanel or other software installed to provide shared hosting - the end user has no more flexibility in scaling than if it were on legacy hardware.

    Yes, 100% uptime is possible and if hardware redundancy were the only thing that effected uptime, cloud hosting as provided today could be the answer. Cloud hosting as provided today still suffers from the same programming issues that have historically caused downtime plus a few new ones.

    Hardware failure has been the cause of a small percentage of issues.

  4. #29
    Quote Originally Posted by Question Everything View Post
    100% uptime (no, it's not impossible, it's just very difficult)
    100% uptime is theoretically impossible, unless you have a finite lifespan for the server which you can measure in retrospect. Even so, just one lost TCP packet means you no longer have 100% uptime. Maybe you mean 99.99999% uptime, but that is entirely different! Sorry to nitpick, but it's misleading when someone states that 100% reliability is an option! Sure, it sounds better when selling to management, but legally you can set yourself up for a world of hurt.
    Last edited by duckribs; 04-20-2011 at 12:57 AM.

  5. #30
    Join Date
    Apr 2011
    Location
    Norfolk Island
    Posts
    2

    Hi all

    Hey all, Fair-minded thought I would respond Hi and thanks for allowing me to till to the forum!

    Nick

  6. #31
    Quote Originally Posted by duckribs View Post
    100% uptime is theoretically impossible, unless you have a finite lifespan for the server which you can measure in retrospect. Even so, just one lost TCP packet means you no longer have 100% uptime. Maybe you mean 99.99999% uptime, but that is entirely different!
    Do you recall the last time the DNS system went down even for a second? I'd say they've achieved 100% uptime.

    A single lost packet does not equal downtime. Instant failover to more than one system, and preferably at least one system in a different physical location can often do the trick. Then if you have to do maintenance on one system, the fully redundant backup takes over while you shut down the primary system. The more fully redundant backups you have the better this configuration works. You can do the maintenance on the backup either before or after you've done it on the primary server. As the backup will need all the capabilities of the primary system this at least doubles your cost. The trick to achieving 100% uptime is eliminating any single point of failure for the system.

    Quote Originally Posted by duckribs View Post
    Sorry to nitpick, but it's misleading when someone states that 100% reliability is an option! Sure, it sounds better when selling to management, but legally you can set yourself up for a world of hurt.
    Ever heard of an SLA? A good one? The harsher the penalties for downtime the more your clients are willing to pay you to not have their system go down. Of course for that to be economical, you'd have to be skilled enough not to have client systems go down.

    100% uptime won't sound so good to management when you give them a cost estimate. The real question is "is a .001% increase in uptime worth the cost?" And for the vast majority of people the answer is no. Nothing is sure in this world - the sun could not come up tomorrow. Would your 100% uptime system survive WWIII? Probably not, but I'm not counting highly unlikely circumstances far outside your control.

  7. #32
    Join Date
    Jul 2010
    Location
    Netherlands
    Posts
    44
    Quote Originally Posted by duckribs View Post
    100% uptime is theoretically impossible, unless you have a finite lifespan for the server which you can measure in retrospect. Even so, just one lost TCP packet means you no longer have 100% uptime. Maybe you mean 99.99999% uptime, but that is entirely different! Sorry to nitpick, but it's misleading when someone states that 100% reliability is an option! Sure, it sounds better when selling to management, but legally you can set yourself up for a world of hurt.
    *vMotionioning a VM to another node as we speak without an glitch of downtime*

    ('nuff said )
    Pr0jects Web Services
    Managed, secure and fast hosting!
    http://www.pr0jects.nl
    http://www.vpsservers.co

  8. #33
    Join Date
    Mar 2006
    Location
    Johannesburg,South Africa
    Posts
    601
    Quote Originally Posted by Pr0jects View Post
    *vMotionioning a VM to another node as we speak without an glitch of downtime*

    ('nuff said )
    what happens when your entire SAN goes offline (yes, it does happen) or the VMotion server crashes (it runs Windows, right?) / looses power (expired UPS battery blows up / power cord accidentally pulled out / office loose power / circuit breaker blows) or looses connectivity to the VMWare servers (virus / network card failure / network cable unplugged / etc)? Do you simply reconnect to the VMWare server and it carries on?

    How about if the switch between the 2 VMWare server & SAN blows a fuse / port / power socket?

    Or, the new server has a faulty LAN, which you only pickup after the VMotion is complete and the VM now keeps crashing? But you can't move it back since the NIC suddenly drops all network connectivity
    South African Web Hosting - http://www.SoftDux.co.za || SA WebHostingTalk - http://www.webhostingtalk.co.za

  9. #34
    Quote Originally Posted by SoftDux View Post
    what happens when your entire SAN goes offline (yes, it does happen) or the VMotion server crashes (it runs Windows, right?)
    LOL, No.
    Quote Originally Posted by SoftDux View Post
    looses power (expired UPS battery blows up / power cord accidentally pulled out / office loose power / circuit breaker blows) or looses connectivity to the VMWare servers (virus / network card failure / network cable unplugged / etc)? Do you simply reconnect to the VMWare server and it carries on?

    How about if the switch between the 2 VMWare server & SAN blows a fuse / port / power socket?

    Or, the new server has a faulty LAN, which you only pickup after the VMotion is complete and the VM now keeps crashing? But you can't move it back since the NIC suddenly drops all network connectivity
    You'd failover to a remote machine that's been in sync from the beginning. The backup takes over and you don't have any downtime.

  10. #35
    Join Date
    Mar 2006
    Location
    Johannesburg,South Africa
    Posts
    601
    Quote Originally Posted by Question Everything View Post
    LOL, No.

    You'd failover to a remote machine that's been in sync from the beginning. The backup takes over and you don't have any downtime.
    If there's a connection problem between the 2 Hypervisors and SAN, you've have downtime.

    And, if the SAN goes down (yes, even those nice expensive ones go down) the whole system goes down
    South African Web Hosting - http://www.SoftDux.co.za || SA WebHostingTalk - http://www.webhostingtalk.co.za

  11. #36
    Quote Originally Posted by SoftDux View Post
    If there's a connection problem between the 2 Hypervisors and SAN, you've have downtime.

    And, if the SAN goes down (yes, even those nice expensive ones go down) the whole system goes down
    The "no" was in reference to running windows. You can solve the SAN single point of failure with redundant systems being kept in sync, so one breaking wouldn't bring your system down. Expensive, yes, impossible, no.

  12. #37
    you may get the cloud from cloud provider like onapp and amazon.
    you need to have 2 server , one is the complete running server and another server is the backup one which is in standby mode to support the first server when something happen to it.
    It costly to get cloud hosting provider but i believe that it will replace all other hosting one day because it do have lot of benefits.

  13. #38
    I have heard that Cloud Computing, Cloud Hosting & Online Storage by Rackspace Hosting. Mosso is now the Rackspace Cloud. is very good cloud host

  14. #39
    Join Date
    Jun 2010
    Location
    Panama
    Posts
    267
    Quote Originally Posted by DewlanceHosting View Post
    I am new to cloud hosting and I never use this,

    How can I provide this?


    Is I need to use some software like Linux cloud on my server?
    The first step is to choose a cloud software, there is a few like Onapp or Applogic(linux cloud isn't).
    Second step is to really pick a good SAN or 2 SAN working in HA mode.
    All the other hardware is standard stuff.

  15. #40
    Join Date
    Jun 2010
    Location
    Panama
    Posts
    267
    Quote Originally Posted by DewlanceHosting View Post
    I am new to cloud hosting and I never use this,

    How can I provide this?


    Is I need to use some software like Linux cloud on my server?
    If you are new to cloud, you first have to decide if you want to rent a cloud or buy the hardware necessary to build one, at the very basic you will need the following:
    2x servers for hyper visors
    1x servers for control panel
    2x servers for making SANs raid 10 in HA setup
    1x Gigabit switch with jumbo frames / layer 2 /vlan
    Offshore Hosting & High Privacy in Panama
    Cloud Servers & Shared Web Hosting | Daily Backups | 99.9% Uptime
    www.OffshoreRacks.com

  16. #41
    Join Date
    Jan 2005
    Location
    Toronto, Canada
    Posts
    588
    http://blog.backblaze.com/2009/09/01...cloud-storage/ interesting link to look at -- on how to build a cheap cloud --

    i think the best way to go is to use onAPP - the future cpanel for us hosters
    Rebel Networks
    Geek Powered Hosting - The Cloud Made Simple
    SuperHero 24 |7 | 365 Support
    Please Visit us @ www.rebelnetworks.com

  17. #42
    Join Date
    Jun 2008
    Location
    San Diego, CA
    Posts
    604
    Quote Originally Posted by OffshoreRacks View Post
    If you are new to cloud, you first have to decide if you want to rent a cloud or buy the hardware necessary to build one, at the very basic you will need the following:
    2x servers for hyper visors
    1x servers for control panel
    2x servers for making SANs raid 10 in HA setup
    1x Gigabit switch with jumbo frames / layer 2 /vlan
    Not exactly. You can have a full HA Cloud using AppLogic with just 2 servers and a backend switch.
    Brent Presley - brent@innoscale.net
    Innovative Scaling Technologies Inc. - Enterprise Cloud Hosting and Support
    24/7 Dedicated Support, Call us @ 1-888-722-8515
    www.innoscale.net - Ashburn - Dallas - Seattle - Santa Clara - Chicago - Amsterdam

  18. #43
    I don't think you understand the meaning of the word "cloud" then. But cloud is just any obnoxious marketing term anyway, and "cloud in a box" is even worse.

  19. #44
    Join Date
    Jun 2008
    Location
    San Diego, CA
    Posts
    604
    RIGHT.

    I understand cloud far better than you. If you want TRUE HA, with TRUE redundancy, and TRUE scalability, you CAN do that with two servers and AppLogic.


    Is it the BIGGEST and most complex cloud designed? No. But if one of the two servers fails, the other continues where the first left off with ZERO down time.

    It's a TINY cloud, but it is a cloud. And IIRC the question was what are the minimum requirements for deploying a cloud. But what the heck do I know. I just deploy these things (and much larger), daily.
    Brent Presley - brent@innoscale.net
    Innovative Scaling Technologies Inc. - Enterprise Cloud Hosting and Support
    24/7 Dedicated Support, Call us @ 1-888-722-8515
    www.innoscale.net - Ashburn - Dallas - Seattle - Santa Clara - Chicago - Amsterdam

  20. #45
    Join Date
    Jan 2011
    Location
    Canada
    Posts
    938
    At that point it sounds like DRBD. Except...

    Note that for grids with exactly 2 servers, AppLogic requires a 3rd reference server in order to properly support the grid controller HA. By default when a 2.7+ grid is installed or upgraded, the AppLogic installer assigns the ALD server as the reference server for the grid. This setting may be overwritten by the maintainer, see the AppLogic installer documentation about the reference server for more details. The same ALD server may be used as a reference server for all grids on the same backbone.

  21. #46
    Join Date
    Jun 2008
    Location
    San Diego, CA
    Posts
    604
    We have not had this issue on 2.9.9. I can assign the primary server as the grid controller, and the HA server as the backup grid controller. We have never brought our ALDO server into the mix for any of our grids. And we have had customers start at 2 servers with plans to scale up after their business grows.
    Brent Presley - brent@innoscale.net
    Innovative Scaling Technologies Inc. - Enterprise Cloud Hosting and Support
    24/7 Dedicated Support, Call us @ 1-888-722-8515
    www.innoscale.net - Ashburn - Dallas - Seattle - Santa Clara - Chicago - Amsterdam

Page 2 of 2 FirstFirst 12

Similar Threads

  1. Replies: 0
    Last Post: 11-15-2010, 04:05 PM
  2. Replies: 0
    Last Post: 10-13-2010, 01:08 PM
  3. Replies: 0
    Last Post: 09-14-2010, 09:31 PM
  4. Replies: 0
    Last Post: 07-27-2010, 01:40 PM
  5. Replies: 0
    Last Post: 07-19-2010, 12:42 PM

Posting Permissions

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