Results 1 to 19 of 19
  1. #1
    Join Date
    Dec 2000
    Location
    Chicago, IL
    Posts
    565

    Comparing cloud and VPS

    Hey guys,

    I have a web design business and I currently have a VPS with ServInt (I pay $50/mo), but I'm looking into possibly switching to either VPS.net or GigaNet.

    I'm surprised that cloud hosting can be had for so cheap.. isn't this a no brainer to go with VPS.net or GigaNet over a regular VPS solution? I'm assuming that having "cloud" hosting could also be a great selling point for potential hosting clients, right?

    For example, my "Essential VPS" plan costs $50/mo with ServInt and has 50GB storage, 1TB monthly transfer, 768MB guaranteed RAM. VPS.net has a plan for $54/mo which has 30GB storage, 1.1TB monthly transfer, 1.1GB guaranteed RAM, and 1.8GHz dedicated CPU.

    How could this not be a no brainer assuming that both of these companies provide top notch service? I actually get a piece of the CPU guaranteed to me unlike VPS.. and all of the benefits of cloud such as high availability, etc.

    Can someone shed some light?

    Thanks,
    Dan

  2. #2
    Join Date
    Jun 2006
    Posts
    1,767
    Honestly, branding aside, a VPS is a VPS. Cloud hosting may be a selling point for your customers but you'll get the occasional spell of downtime with any provider. I haven't heard many bad things about either of those companies, but consider other options too. A 'non-cloud' VPS can offer the same reliability and speed.

  3. #3
    Join Date
    Dec 2004
    Location
    New York, NY
    Posts
    10,710
    On the ServInt VPS, although you don't have guaranteed CPU, you likely have the capability of bursting to a higher amount of CPU than what is dedicated to you on a cloud server at VPS.net. Also, the ServInt VPS is fully managed, I don't believe vps.net offers management, at least not at that price.

    Finally, a cloud adds a lot of complexity over standard virtualization (mainly a SAN instead of local storage). Yes a cloud is often designed to be redundant, etc. but you still hear about many failures regardless, and that has to do with greater complexity in the infrastructure. On the other hand ServInt has been doing VPS hosting for years now, their setup is mature and with the hardware that is typically running a VPS node you can still have very high reliability and great performance.

    So, overall, if you are happy with your current VPS at ServInt, I wouldn't necessarily leave it for a cloud server just because of a better perceived value for the price.

  4. #4
    Join Date
    Jan 2004
    Location
    Chicago
    Posts
    984
    If you're on a budget, and your existing VPS solution has been reliable and no complaints.. I'd stick with that.

    Most cloud providers at the moment while they look very good on paper with all the features still have a fair amount of issues with their systems and downtime that I haven't seen with tried and true VPS systems or enterprise virtualization products.

    Pricing is also a tossup, there's a few with good pricing, but tacking on all the little extras like IP's, control panels, metered bandwidth, backup solutions, etc.. all add up quickly. Most of those prices aren't disclosed up front until you get into the backend customer panel and start configuring your server, something I think providers need to be more upfront about.

    If you are set on trying out cloud services, at least take advantage of the hourly and daily service options to evaluate them for a week or two, or request a trial credit.

  5. #5
    Join Date
    Jul 2006
    Location
    Lake Zurich, IL
    Posts
    436
    I would add that, depending on your knowledge of servers, storage, networks, etc., it's good to find out what your VPS or cloud server is actually running on, including physical servers, storage, and hypervisor, and what redundancy is used at every layer.

    There is a misconception of what physical servers are used by cloud providers (e.g. EC2, LiquidWeb, Rackspace Cloud, etc.). People see "cloud" and think reliability. This is typically the opposite case, where cloud servers are designed for scale-out and quick provisioning, where the application handles failure, not the infrastructure, and failure is assumed to be a normal thing. For more common workloads, you may actually have better luck with a VPS running on high quality hardware with redundant parts, or a cloud service that concentrates on highly-available environments.

    Eric
    Genesis Hosting Solutions, LLC (genesishosting.com)
    Genesis Public Cloud - No Compromise On-Demand OpenStack infrastructure
    Genesis VMs - Easy to provision single VMs on our Genesis Public Cloud
    Compare us against others at vpsbenchmarks.com!

  6. #6
    I still prefer VPS over cloud hosting, and for our secure customers, we put them on their own VPS system rather than a shared hosting system.

  7. #7
    Join Date
    Jun 2006
    Location
    Lansing, Michigan
    Posts
    649
    Quote Originally Posted by layer0 View Post

    Finally, a cloud adds a lot of complexity over standard virtualization (mainly a SAN instead of local storage).
    Tiny clarification, not all Cloud products are SAN based. Storm On Demand, for instance, uses local storage. (I suppose we could debate the term "cloud" though since that is also a point of debate still)
    Travis Stoliker
    Liquid Web - Dedicated Hosting with Heroic Support
    StormOnDemand - Flexible Cloud Hosting Infrastructure
    1-800-580-4985 | Twitter: @liquidweb | @StormOnDemand

  8. #8
    Join Date
    Jun 2006
    Location
    Lansing, Michigan
    Posts
    649
    Quote Originally Posted by erickmiller View Post

    There is a misconception of what physical servers are used by cloud providers (e.g. EC2, LiquidWeb, Rackspace Cloud, etc.). People see "cloud" and think reliability. This is typically the opposite case, where cloud servers are designed for scale-out and quick provisioning, where the application handles failure, not the infrastructure, and failure is assumed to be a normal thing. For more common workloads, you may actually have better luck with a VPS running on high quality hardware with redundant parts, or a cloud service that concentrates on highly-available environments.

    Eric
    If I understand you correctly, there are a couple exceptions to this. A few providers offer "Cloud Features" on wholly allocated physical servers and the customer can select specific sever configurations. A few companies provide this but since you specifically mentioned Liquid Web, I would like to point out that Liquid Web offers this as "Storm On Demand Bare Metal Servers".

    Again, we could certainly debate the term "cloud" since that is still being debated but I'll leave that conversation for another time
    Travis Stoliker
    Liquid Web - Dedicated Hosting with Heroic Support
    StormOnDemand - Flexible Cloud Hosting Infrastructure
    1-800-580-4985 | Twitter: @liquidweb | @StormOnDemand

  9. #9
    Join Date
    Dec 2005
    Location
    I'm Lost...Help
    Posts
    895
    I have been testing many different options lately to find the best solution and there is what I found from my experience.

    1. Openvz - I think is a decent system overall. Speed and performance are not bad, but I would not recommend for anything to resource intensive..

    2. Cloud - I have tried Vps.net and now Gigenet oh their cloud based systems and I was not impressed one bit. They seemed very slow and unresponsive. I don't know if it is the OnApp system of the SAN drives. Either way I was not happy with the testing.

    3. I just picked up a nice Xen server from Quickweb and I am amazed at the speed and performance. This thing really feels like a dedicated server when you are working with. Although I just got it, as soon as I started performing actions on the server and loading websites I was shocked at how snappy it was.

    So after all this I have to say I am not impressed with the cloud systems at all. If anything try a nice Xen server and I think you will be overly impressed.
    Kevin Kopp - MonsterMegs Business Class Hosting Services
    Pure SSD Powered Shared, Reseller, and Enterprise Hosting Solutions
    US & NL Locations :: [US] PhoenixNAP | [NL] EvoSwitch Datacenters

  10. #10
    Join Date
    Jul 2006
    Location
    Lake Zurich, IL
    Posts
    436
    Quote Originally Posted by LiquidWebTravis View Post
    If I understand you correctly, there are a couple exceptions to this. A few providers offer "Cloud Features" on wholly allocated physical servers and the customer can select specific sever configurations. A few companies provide this but since you specifically mentioned Liquid Web, I would like to point out that Liquid Web offers this as "Storm On Demand Bare Metal Servers".

    Again, we could certainly debate the term "cloud" since that is still being debated but I'll leave that conversation for another time
    Correct, that there are exceptions. What we see is most people viewing cloud infrastructure as a magic solution to provide highly-available services, which "can" be done at a higher layer with the right software solution, but the majority of software isn't designed to scale horizontally in all respects, much less handle hardware failures gracefully and un-noticeably.

    Larger cloud providers are competing with the EC2 model which wasn't designed for reliability at the individual physical server level, but is rather a model allowing for extreme scale-out with low-cost commodity systems (generally equating to as cheaply put together as possible to keep the price as low as possible), while providing the most flexible (and thus variable) pricing model possible. Customers pay a premium for this scalability and pricing flexibility compared to longer-term rented dedicated systems with higher-quality components. EC2 has a large number of unused servers, waiting for spikes, that everyone is paying for in the end.

    With any provider, it's simply important to understand what the physical and logical components consist of, where the points of failure are, and that it is a good fit for a particular application. Unfortunately, many details are hidden by providers, which doesn't make shopping easy. Cloud services, such as EC2 and their competition, don't qualify as a very good platform to run single workloads (not distributed) due to the inherent nature of the design.

    Eric
    Genesis Hosting Solutions, LLC (genesishosting.com)
    Genesis Public Cloud - No Compromise On-Demand OpenStack infrastructure
    Genesis VMs - Easy to provision single VMs on our Genesis Public Cloud
    Compare us against others at vpsbenchmarks.com!

  11. #11
    Join Date
    Jan 2010
    Posts
    1,030
    Quote Originally Posted by erickmiller View Post
    Correct, that there are exceptions. What we see is most people viewing cloud infrastructure as a magic solution to provide highly-available services, which "can" be done at a higher layer with the right software solution, but the majority of software isn't designed to scale horizontally in all respects, much less handle hardware failures gracefully and un-noticeably.
    True Cloud software is designed to scale horizontally, and vertically, and handle hardware failures gracefully. That's a major point of Cloud. You may be comparing virtualization platforms.. not Cloud. Virtualization is just a component of Cloud like an engine is a component of a car.

    Quote Originally Posted by erickmiller View Post
    Larger cloud providers are competing with the EC2 model which wasn't designed for reliability at the individual physical server level, but is rather a model allowing for extreme scale-out with low-cost commodity systems (generally equating to as cheaply put together as possible to keep the price as low as possible), while providing the most flexible (and thus variable) pricing model possible. Customers pay a premium for this scalability and pricing flexibility compared to longer-term rented dedicated systems with higher-quality components. EC2 has a large number of unused servers, waiting for spikes, that everyone is paying for in the end.
    That's also not true. Most Cloud providers that I know personally spend significantly more in their infrastructure than providers offering traditional hosting services which many use "pc level" hardware, low-end networking gear, and lacking redundancy in the environment from everything from power and networking, to components in each server. Cloud does have some overhead hardware but I would argue it's less than non-cloud. Why? We keep a large inventory of various hardware for our traditional platforms, far more than that on the Cloud as we use the same hardware for each Cloud.

    Because of the dynamics in the environment, we can fill each server to 100% as when a customer needs more resources they will automatically self-migrate to another server with more resources. Traditional VPS customers are tied to the server and when they want to upgrade, they usually have to migrate to another server because the provider has it full. If they don't have it full, then they are wasting resources more than Cloud does waiting for users to grow.

    Quote Originally Posted by erickmiller View Post
    With any provider, it's simply important to understand what the physical and logical components consist of, where the points of failure are, and that it is a good fit for a particular application. Unfortunately, many details are hidden by providers, which doesn't make shopping easy. Cloud services, such as EC2 and their competition, don't qualify as a very good platform to run single workloads (not distributed) due to the inherent nature of the design.

    Eric
    That looked good until you got to the part of making the assumption that Cloud does not quality as a very good platform to run single workloads.

    Cloud is simply a better way to do things. There is no longer an argument that is "Cloud" alright. It's beyond early majority and now into late majority. The US Government is adopting Cloud in every organization (see Cloud First policy passed last month), Small business is adopting it, Medium sized business is adopting it.. and Enterprises are adopting it. The calls we get are no longer if "Cloud" is right for their business, but instead how can they adopt it in their business properly as there are many ways to provide Cloud services.

  12. #12
    Ok it is all about cloud hosting IMO don't get me wrong vps is ok but not as good as cloud vps it is truly amazing I wish I had one I can only dream.. I want a cloud dedi really bad!

  13. #13
    Yes, a cloud hosting can be very effective than VPS. But sometimes as they advertise, real thing happens something different, so need to do a complete check before hopping to a cloud service provider, must do a search of negative feedback's at all at prominent communities like WHT <<snipped>>. I hope this servers your answer.
    Last edited by bear; 01-20-2011 at 01:37 PM.

  14. #14
    Join Date
    Jan 2011
    Location
    Houston
    Posts
    61
    You will have more downtime with a VPS solution and some more restrictions. If you go with a cloud VMware provider you will have more redundant enterprise failover options. VPS is good for basic hosting needs. Cloud/VMware is more for a corporate infrastructure (multiple servers)or for a highly available web server. VMware is amazing if you have a SAN, you will need one to do the cool stuff. You can even failover data centers. Also with VMware you can more control over each vm. If your infrastructure is designed right SATA drives in a SAN should be fine. A little off topic but wanted to point out some things.
    Last edited by FastPCNet; 01-20-2011 at 04:13 PM. Reason: clarification
    FastPCNet Online Backup, Houston Colocation and Data Center Services
    Call us at 877 614 host for a quick quote

  15. #15
    Join Date
    Jul 2006
    Location
    Lake Zurich, IL
    Posts
    436
    Quote Originally Posted by CloudWeb View Post
    True Cloud software is designed to scale horizontally, and vertically, and handle hardware failures gracefully. That's a major point of Cloud. You may be comparing virtualization platforms.. not Cloud. Virtualization is just a component of Cloud like an engine is a component of a car.
    Please name by example (e.g. "True Cloud software").

    Quote Originally Posted by CloudWeb View Post
    That's also not true. Most Cloud providers that I know personally spend significantly more in their infrastructure than providers offering traditional hosting services which many use "pc level" hardware, low-end networking gear, and lacking redundancy in the environment from everything from power and networking, to components in each server. Cloud does have some overhead hardware but I would argue it's less than non-cloud. Why? We keep a large inventory of various hardware for our traditional platforms, far more than that on the Cloud as we use the same hardware for each Cloud.
    Please name by example (e.g. "Most Cloud providers"), other than yourself.

    Quote Originally Posted by CloudWeb View Post
    Because of the dynamics in the environment, we can fill each server to 100% as when a customer needs more resources they will automatically self-migrate to another server with more resources.
    Please clarify. Not sure why you would want to fill a server at all. Why not spread VMs across all nodes as evenly as possible?

    Quote Originally Posted by CloudWeb View Post
    That looked good until you got to the part of making the assumption that Cloud does not quality as a very good platform to run single workloads.
    In common cloud systems, which is what I mentioned in my response ("Cloud services, such as EC2 and their competition") where shared storage and replicated storage is not used (EC2, LiquidWeb StormOnDemand, Rackspace Cloud, and other upcoming vendors such as BrightBox), loss of data will occur, with significant time to recover when (not if) a single physical server fails. A backup can be restored to a good node to bring that workload back online, but only at the point in time when the backup occurred. That's what I mean by "not a very good platform to run single workloads"... potential loss of data, large time to recover to the last backup. Obviously, recovery time will depend on the backup method used, but having to manually deal with this at all isn't good for any single workload.

    I'm I'm missing the boat, please describe step by step the recovery methods used during a failure in one of the systems mentioned above and why that's better than an environment with shared storage and automatic node fail-over (non-EC2-like environments).

    Also, if I'm mistaken about local versus shared storage in any of the providers mentioned, feel free to correct me.

    Quote Originally Posted by CloudWeb View Post
    Cloud is simply a better way to do things. There is no longer an argument that is "Cloud" alright. It's beyond early majority and now into late majority. The US Government is adopting Cloud in every organization (see Cloud First policy passed last month), Small business is adopting it, Medium sized business is adopting it.. and Enterprises are adopting it. The calls we get are no longer if "Cloud" is right for their business, but instead how can they adopt it in their business properly as there are many ways to provide Cloud services.
    Under the right conditions, 3rd-party computing services are the way to go. That doesn't mean all cloud services are a better approach for "everything".

    Eric
    Genesis Hosting Solutions, LLC (genesishosting.com)
    Genesis Public Cloud - No Compromise On-Demand OpenStack infrastructure
    Genesis VMs - Easy to provision single VMs on our Genesis Public Cloud
    Compare us against others at vpsbenchmarks.com!

  16. #16
    Join Date
    Jan 2010
    Posts
    1,030
    Quote Originally Posted by erickmiller View Post
    Please name by example (e.g. "True Cloud software").



    Please name by example (e.g. "Most Cloud providers"), other than yourself.



    Please clarify. Not sure why you would want to fill a server at all. Why not spread VMs across all nodes as evenly as possible?



    In common cloud systems, which is what I mentioned in my response ("Cloud services, such as EC2 and their competition") where shared storage and replicated storage is not used (EC2, LiquidWeb StormOnDemand, Rackspace Cloud, and other upcoming vendors such as BrightBox), loss of data will occur, with significant time to recover when (not if) a single physical server fails. A backup can be restored to a good node to bring that workload back online, but only at the point in time when the backup occurred. That's what I mean by "not a very good platform to run single workloads"... potential loss of data, large time to recover to the last backup. Obviously, recovery time will depend on the backup method used, but having to manually deal with this at all isn't good for any single workload.

    I'm I'm missing the boat, please describe step by step the recovery methods used during a failure in one of the systems mentioned above and why that's better than an environment with shared storage and automatic node fail-over (non-EC2-like environments).

    Also, if I'm mistaken about local versus shared storage in any of the providers mentioned, feel free to correct me.



    Under the right conditions, 3rd-party computing services are the way to go. That doesn't mean all cloud services are a better approach for "everything".

    Eric
    1) CA/3tera AppLogic.

    2) It would be unprofessional for me to list by name on a public forum the name of associate companies without their explicit permission. It could cause an illusion of a business partnership where one may or may not exist. However, much of infrastructure information is public though as most providers will gladly talk about their business as they have invested heavily into it.

    3) We do spread them equally, but the point is they can fill to 100% dynamically without causing restrictions.

    4) You are right about some providers not providing redundancy of data and that is a true shame as IMHO those providers who do not have redundant data, are not true cloud. I think this is absolutely imperative to Cloud. For some providers they think "centralized storage" is enough.. I think it needs to be "centralized and redundant storage". In the case of those using AppLogic, we utilize local storage on each server that is then pooled into a virtual network SAN if you will.. every instance is replicated and the users VM is not tied to any physical server. Data can be accessed on either of the storage locations from local or remote servers. Step by step recovery methods will vary based on the application and the customer method but replication is provided inherently in the infrastructure. Think of it like a network based RAID1. Obviously not a replacement for backups.. but if a server crashes the data does not.

    Cloud can be private, it does not need to be outsourced to third parties. We're building private Clouds for people with as little as two servers (N+1 for redundancy).

    5)

  17. #17
    Join Date
    May 2007
    Posts
    451
    Quote Originally Posted by CloudWeb View Post
    1) CA/3tera AppLogic.

    2) It would be unprofessional for me to list by name on a public forum the name of associate companies without their explicit permission. It could cause an illusion of a business partnership where one may or may not exist. However, much of infrastructure information is public though as most providers will gladly talk about their business as they have invested heavily into it.

    3) We do spread them equally, but the point is they can fill to 100% dynamically without causing restrictions.

    4) You are right about some providers not providing redundancy of data and that is a true shame as IMHO those providers who do not have redundant data, are not true cloud. I think this is absolutely imperative to Cloud. For some providers they think "centralized storage" is enough.. I think it needs to be "centralized and redundant storage". In the case of those using AppLogic, we utilize local storage on each server that is then pooled into a virtual network SAN if you will.. every instance is replicated and the users VM is not tied to any physical server. Data can be accessed on either of the storage locations from local or remote servers. Step by step recovery methods will vary based on the application and the customer method but replication is provided inherently in the infrastructure. Think of it like a network based RAID1. Obviously not a replacement for backups.. but if a server crashes the data does not.

    Cloud can be private, it does not need to be outsourced to third parties. We're building private Clouds for people with as little as two servers (N+1 for redundancy).

    5)
    I couldn't agree more.

    To be cloud I think you need to have 100% diverse and redundant data storage. In Applogics case, I can loose a back plane on my cloud (1 server) and still retain data integrity. On most of our stuff we like to keep N+2, and we are also able to mirror volumes across data centers, for DR. While we have yet to use it in the 3 years we've been offering Applogic.

    Thats one of the reasons i think Applogic is going to outsell the rest. The ability to not need a SAN array, and to also provide 100% private clouds down to storage, and backend networking.

    Thanks,
    Michael Wallace - michael@innoscale.net
    Innovative Scaling Technologies Inc. - A Cloud Service Provider
    24/7 Support, Call us @ 1-307-200-4880
    www.innoscale.net - Seattle, Silicon Valley, Dallas, Chicago, Washington D.C., and Europe

  18. #18
    Join Date
    Jul 2006
    Location
    Lake Zurich, IL
    Posts
    436
    Quote Originally Posted by CloudWeb View Post
    1) CA/3tera AppLogic.
    I personally don't know much about 3tera, but I've seen the name often. I'll have to read more about it.

    Quote Originally Posted by CloudWeb View Post
    2) It would be unprofessional for me to list by name on a public forum the name of associate companies without their explicit permission. It could cause an illusion of a business partnership where one may or may not exist. However, much of infrastructure information is public though as most providers will gladly talk about their business as they have invested heavily into it.
    I think talking about the technology that another vendor uses isn't necessarily unprofessional. Bashing is definitely unprofessional. In fact, discussing what others are using (as long as it is an accurate, technical discussion with back-able facts) should be encouraged. What good is a forum like this if we can't? We certainly can't talk about our own technology (from what I've seen on WHT unfortunately) since we'll be considered self-promoting.

    Quote Originally Posted by CloudWeb View Post
    3) We do spread them equally, but the point is they can fill to 100% dynamically without causing restrictions.
    I thought so, just wanted to be sure. The only reason I've seen for squeezing more VMs onto as few nodes as possible is if power utilization could be reduced. I have yet to see a facility that meters, though, unfortunately, otherwise we would probably use this technology.

    Quote Originally Posted by CloudWeb View Post
    4) You are right about some providers not providing redundancy of data and that is a true shame as IMHO those providers who do not have redundant data, are not true cloud.
    I completely agree, which was my point of referring to EC2-like clouds, which most consider to be the industry standard. EC2 has its place, but again, for single workloads, it's not very good, because of the lack of redundancy, and reliance on the application to provide its own redundancy and resiliency to hardware failures. That's all I was trying to say.

    Quote Originally Posted by CloudWeb View Post
    In the case of those using AppLogic, we utilize local storage on each server that is then pooled into a virtual network SAN if you will.. every instance is replicated and the users VM is not tied to any physical server. Data can be accessed on either of the storage locations from local or remote servers.
    So it sounds like virtual SAN solutions similar to LeftHand Networks (now HP-owned), StorMagic, and Seanodes.

    Quote Originally Posted by CloudWeb View Post
    Step by step recovery methods will vary based on the application and the customer method but replication is provided inherently in the infrastructure. Think of it like a network based RAID1. Obviously not a replacement for backups.. but if a server crashes the data does not.
    This is definitely the right way to do it... with redundant storage. What I was referring to is the EC2-like systems, where there isn't redundancy, and thus restoration after a hardware failure would be the only solution (for single workloads).

    Eric
    Genesis Hosting Solutions, LLC (genesishosting.com)
    Genesis Public Cloud - No Compromise On-Demand OpenStack infrastructure
    Genesis VMs - Easy to provision single VMs on our Genesis Public Cloud
    Compare us against others at vpsbenchmarks.com!

  19. #19
    Hi,

    Thanks for sharing the information. It is really interesting. I hope it will help me in my seminars.

Similar Threads

  1. Replies: 1
    Last Post: 04-18-2010, 04:24 PM
  2. Replies: 0
    Last Post: 02-15-2010, 02:20 PM
  3. Replies: 1
    Last Post: 10-29-2009, 08:06 PM
  4. Comparing my VPS with yours..
    By 5wayshost in forum VPS Hosting
    Replies: 12
    Last Post: 11-03-2008, 11:03 AM
  5. Comparing VPS Hosting Providers
    By snapmetrics in forum Other Reviews
    Replies: 20
    Last Post: 07-05-2006, 03:08 AM

Posting Permissions

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