>Thanks. Yes, these are the things we're looking for. Can you offer any specifics, examples or details?
Syncthing has proven to be efficient for file synchronization from a data center to a local environment. You can control its thresold and related settings so it wouldn't eat all the bandwidth, a small setting most would not do at first realizing it later when something stops to respond.
If you really need to control everything the same way, with maximum control, than IAAS could be considered. It does require more technical expertise. This is one reason why many would not use it. They would go for PAAS or SAAS, also depending on their needs. The cost wouldn't vary that much. I have analyzed IAAS pricing for specific needs compared to Azure, and it turned out at the end it was costly using Azure. As a whole, cloud providers would try to have packages which could offer as much as possible the same flexibility you used to have from your local infrastructure. Thought impossible at first, things have evolved a lot since a few years, and we are getting extremely close to that level now.
The thing that bothers me more is IOPS. Thus, you might want to check on that with your cloud provider to see about related contexts. When you work with your local infrastructure, you usually have no dependency at the hardware level or related restrictions. In the cloud, IOPS is usually part of the monthly package and would leverage across your setup. For your SQL Server, split everything on various disks so you can avoid IOPS related side effects. MDF goes on one disk, LDF on another, OS on another, and backup on another. Try to calculate a proper IOPS setting that would be ok for everything.
I have been told that a dedicated virtual setup would not be affected by the traffic other clients on the same host would do. No one has been able to guarantee that so far as it is extremely difficult to foresee. For me, I do not want another client to cause latency on my setup. It is normal we all think like that. Thus, if you run into a situation that you have no traffic on your network, but you have a random 15-second delay just appearing like that, then this is a red falg. This is also another important thing to verify with your provider.
Those things, if you only deal with them once you are in production, you won't get the same support. You're better off negotiating all that at the very first.
You may want to check for SLA related clauses. When something goes down, which would be out of your control, you are eligible for a credit. Most important, you want to have security in mind that the issue will be addressed properly.
As for firewall, everyone wants a dedicated firewall. Most would adjust a little bit when they receive information that it will cost more. Just normal after all as I wouldn't want someone else to manage my own firewall. Depending on IAAS vs PAAS, for example, you might have different level of firewall services.
Most important, scalability. You would probably want to increase RAM as required, in no time, when a situation occurs. And, this is great within those environments. You can increase of decrease as you see fit. You just do it from a host interface and that's it. Hoever, it might require a remote startup of the VM. The same goes with storage and CPU. So, you no longer need to think too far in time as you know you have that flexibility. Thus, you do not have to buy ahead and have everything in the server box for years to come. You can then start with a setup that would be ok for six months and increase gradually. If that is the case, then it is a good sign as it means the business is going well. :)
As you can see, some of that allows to save money, not buying too much in one shot at the very first. This new tendency allows businesses to concentrate more on the business operations than managing the hardware, which has been given a few nightmares to many since a generation.
One thing, in such environment, document everything, each provider regular scheduled maintenance, each anomaly, what was done for their resolution, and related issues. Within such, we delegate some responsibilities to an entity we can now define as a business partner, sort of, thus, we are no longer in full control of everything.
If you go with PAAS and solutions like Azure, make sure you have a package that is scalable. Up to last year, there was still a lot of that that was not able to scale after you purchase it. In such, there wouldn't much advantage to go in virtual mode if we have to negotiate the package in the middle of a term.
If you have clients, this is also a good seller. Not everyone like, nowadays, to have data which sits in the office itself. ISP have power and Internet redundancy infrastructure, one reason that management likes to make that switch, finally.
If you go IAAS, don't hesitate to ask them to pay a small visit. I wouldn't have recommended to sign with someone who wouldn't have allowed us to visit their facility. This provides a sentiment of security and trust. After all, this is a required concept if we are to send our data over there.
The longer you sign, the bigger the save. At first, many are relunctant to sign a multi-year deal. You might save close to 33% going to three years on your first term instead of just one year. This applies for IAAS and probably PAAS I would guess.
There's a lot of concerns and other points to talk about when making such a move. A good project management would be a great asset in this case.
One thing for sure, this is the future. So, you know you are taking a good decision moving forward into such virtual world.
