4 replies to this topic

[Bluesplinter]

Alrighty, well now then... here I am, a bright-eyed young student on his first day of college, eager to learn all about "the cloud". Assuming I know nothing (which is pretty darn close to the truth), what can you tell this hillbilly about the cloud as it relates to Cartika?

1. What separates your cloud server from a traditional VPS? Is it just the instant scalability of ram/hd/cpu, or are there also infrastructure differences that distinguish the two?

2. What does self-healing actually mean? Is there some monitoring software that sees a cloud server go down, and instantly bring it up elsewhere in "the cloud"? Or do I have to notice it (either from my own monitoring, or an angry client report), and request a "healing" via ticket? Since you use the term "self" -healing, I'm assuming this is automatically monitored...? How long does this process take? Seconds, minutes? Is any data lost in the process? Do we have to restore from CDP once the new instance is up, or does it read from the original SAN?

3. Is a cloud server any more HA than your regular cluster? You already have 99.9+% uptime on your regular cluster... can we expect to beat that with a cloud server?

4. What kind of interface will we use to control our server instances? I'm not talking about the hosting panel, but rather where we would go to boost ram, cpu, etc.

5. What speed and flavor is your cpu node?

6. Do you monitor new releases of OS and hosting stack (LAMP, etc) software? On a new mysql release (for example), do you notify us and let us decide what to do, or do we need to watch for these various releases, and then request updates as needed? Obviously I don't want to be on the bleeding edge... I like for new releases to be vetted for a bit before upgrading, but at the same time, if an urgent security hole is discovered and patched, waiting would be bad. :)

7. Any rough ETA?

That should do for a start :)
Steve

[CH-Andrew]

Hello Steve and sorry for the delayed response here. These "CartikaCloud" forums are brand new and apparently we did not setup alerting properly. So, I did not see this until a staff member pointed it out to me today... I will answer your questions inline below..

Firstly, you are a little too quick on the draw here :) we have not yet launched, so, people might not be exactly sure what you are talking about.

Cartika is launching its new Cloud solution SOON - stay tuned. The website will be www.cartikacloud.com (you will not see the site there at as we have not launched yet :) ) - unfortunately, until launched, you will just see the place holder page (default hsphere page)

Bluesplinter said:

1. What separates your cloud server from a traditional VPS? Is it just the instant scalability of ram/hd/cpu, or are there also infrastructure differences that distinguish the two?

Well, the cloud is still fuzzy, and different people are marketing different things (including us with our existing HAL offering). Some people simply have a fancy billing control panel wrapped about VPS offerings and are selling the solution as a Cloud.

Our solution is closer to an Amazon Cloud then it is to something like our HAL offering or Slicehost/Rackspacecloud offering.

The basic differentiators are:

• Self Healing Hardware where Cloud Servers are instantly brought up on alternate hardware in case of failure
  
• Ability to seamlessly migrate Cloud Servers between physical nodes if a customer needs more resources then is available on their particular physical node (all automatic and instant)
  
• Ability for customers to add or remove hard resources on demand (ie CPU and RAM) and pay by the day or month for the additional resources when used
  
• Cloud Storage built on NetApp systems. Scalable, redundant, proven platforms vs others who may be using homegrown SAN solutions, etc
  
• A dedicated Cloud Network that is simply redundant and scalable - no other way to look at this. It is doubtful that many (or ANY) have built out their network for their Cloud Services in this manner. I am fairly confident about this, since Citrix utilities was unable to work within our network as they are used to much more basic configurations and had not seen or thought of how to deal with, the network we have configured. Each server has 8 gig connections, routing to redundant Cisco switching and routing depending on specific requirements (LAN, WAN, Storage, Backups). It took some time to get everything to work exactly how we wanted it to.
  
• For the Cloud Storage Network we are using a combination of bonded GIG Ethernet and 10G connections depending on the specific service and our specific requirements. The great thing with NetApp is that we can address specific bottlenecks as they come up. No one really knows how these things will scale. What happens when we have 1000s of cloud servers loaded up. Will IO be the issue? Will throughput be the issue?, etc.. the reason we made the NetApp decision is because of the true Enterprise grade we are dealing with. There are no limitations. Their storage systems provide extensive graphing and data analytics. We can proactively determine where we are going to have bottlenecks in the future and proactively modify the systems to increase capacity to those areas, without any fear of running into service affecting issues and learning via trial and error. We felt this was critical and well worth the extra expenditure in this hardware. As far as I know, none of our competitors are using anything this advanced and robust to back end their cloud offerings..

I am sure there is more, but, those are the key points that would probably interest most people..

Quote

2. What does self-healing actually mean? Is there some monitoring software that sees a cloud server go down, and instantly bring it up elsewhere in "the cloud"? Or do I have to notice it (either from my own monitoring, or an angry client report), and request a "healing" via ticket? Since you use the term "self" -healing, I'm assuming this is automatically monitored...? How long does this process take? Seconds, minutes? Is any data lost in the process? Do we have to restore from CDP once the new instance is up, or does it read from the original SAN?

The servers in a Cloud formation are essentially just raw resources. All data is stored on the storage systems (NetApp in this case). It does not matter which physical node your cloud server is located on. The configs for all cloud servers exist on all physical nodes. If 1 server fails, requests start automatically being served from available servers with the available resources. This is all instantaneous. And since all data is not located locally on the actual physical node, it is completely transparent from that perspective..

Quote

3. Is a cloud server any more HA than your regular cluster? You already have 99.9+% uptime on your regular cluster... can we expect to beat that with a cloud server?

Cloud Servers essentially remove downtime associated with hardware failures and with capacity issues (as you can add more capacity on demand). So, yes, I would expect to see better uptime metrics over the long haul with a Cloud Server vs our standard shared cluster.

Quote

4. What kind of interface will we use to control our server instances? I'm not talking about the hosting panel, but rather where we would go to boost ram, cpu, etc.

We have built an interface for this where all of your servers are listed, you can access console, add or remove resources, manage your billing, etc..

Quote

5. What speed and flavor is your cpu node?

All CPUs are Intel Nehalem E5520's running at 2.26 Ghz

Quote

6. Do you monitor new releases of OS and hosting stack (LAMP, etc) software? On a new mysql release (for example), do you notify us and let us decide what to do, or do we need to watch for these various releases, and then request updates as needed? Obviously I don't want to be on the bleeding edge... I like for new releases to be vetted for a bit before upgrading, but at the same time, if an urgent security hole is discovered and patched, waiting would be bad. :)

This depends, For Self Managed Cloud Servers, people maintain these sorts of things themselves. Our managed solutions will continue to run as they always have. We will update, monitor, backup, etc any managed systems just like we do today with shared, reseller, HAL, VPS or dedicated server customers

Quote

7. Any rough ETA?

Very very shortly..

We are looking to move a couple of our HAL customers over to the Cloud now. If you have any interest, please let us know and we can probably start that process today or tomorrow.

All HAL customers will receive a new, fully managed cloud server for the exact same price they are currently paying for HAL (significant discount over our go to market price for a managed Cloud Server). So, if it is something you are interested in, please contact our support department and we can get the process rolling

thanks for your interest and your comments/questions.. hopefully the answers are useful and explain things better. More will become apparent once the site is formally launched (likely this weekend or early next week)

[CH-Andrew]

Hello Steve,

since we are so close to launching, I figure we might as well give people a sneak peak to the website

www.cartikacloud.com/index.php

Now, please be warned, this site is not finished yet and nothing is absolutely finalized yet as far as pricing, etc goes..

a couple of real interesting stuff to add as well, including our partnership with cloudlinux.com (more to come :) )

[Bluesplinter]

Hey Andrew,

Sounds awesome!

Thanks for all the details... not only have they whetted my appetite, I am positively drooling in anticipation . I originally had some confusion between your new cloud and a traditional VPS, but you have now cleared that up admirably. :)

And as for jumping the gun, hey, you put the ball in play with these empty forums! ;)

I'm on my way to the support crew now!

Take care,
Steve

[CH-Andrew]

Bluesplinter said:

And as for jumping the gun, hey, you put the ball in play with these empty forums! ;)

LOL - sorry about that... guess we should have waited to launch the forums.. thanks for your comments and enjoy your new cloud server.. please make sure you come back and tell us all how great it is :)