Solved Hyper V replica VS Veeam B&R Replica.
-
@openit said in Hyper V replica VS Veeam B&R Replica.:
@scottalanmiller said in Hyper V replica VS Veeam B&R Replica.:
@Dashrender said in Hyper V replica VS Veeam B&R Replica.:
@openit said in Hyper V replica VS Veeam B&R Replica.:
@scottalanmiller said in Hyper V replica VS Veeam B&R Replica.:
@openit said in Hyper V replica VS Veeam B&R Replica.:
@JaredBusch said in Hyper V replica VS Veeam B&R Replica.:
@openit said in Hyper V replica VS Veeam B&R Replica.:
This is my future plan to setup Windows Server Redundancy ( DC+File Server).
Go back and decide if you need redundancy from a business point of view.
I believe, Yes.
For Management : Minimized Downtime
For IT : Peace of Mind
That's never how you should look at it.
For business: Whatever makes the money
IT: Whatever is good for the businessMaybe I was wrong at "For IT: Peace of Mind"
As I said, our end-users depends a lot on File Server. If File Server is not available, they can't do their work almost (lets say 70%). So if they don't do work for long, their projects will get delayed, and for sure it will effect production.
Of course. But this is a business concern, not a technical one. If the business says they can afford 1 day of full downtime with few issues, spending money to reduce much below that mark is directly against IT's mission.
Right. The only thing that IT should be doing is getting the numbers from the business and supplying the cost of what it takes to mitigate it and how well that mitigation should work. This is 100%, no exceptions, a math problem and if no math is being done, no one has even started evaluating what the need will be.
Okay. I will take help from Management to do maths about Redundancy requirement.
Let's assume, Server Redundancy is necessary after doing maths, could you please advice which one option for Server Redundancy is better ?
Well that would involve other pieces. As stated before, if you are going to purchase Veeam for backups, then Veeam is better for the extra features such as alerting.
If you go with some other method of backup, that does not have Replication tools alos, then just user Hyper-V replication. it works well.
-
Yup, you'd be back to something like Hyper-V Replication (the simple way) or using a more complex tool like Starwind. Either way, you have a free storage clustering option.
-
@scottalanmiller said in Hyper V replica VS Veeam B&R Replica.:
Yup, you'd be back to something like Hyper-V Replication (the simple way) or using a more complex tool like Starwind. Either way, you have a free storage clustering option.
Hyper-V replica is not clustering.
-
@JaredBusch said in Hyper V replica VS Veeam B&R Replica.:
@scottalanmiller said in Hyper V replica VS Veeam B&R Replica.:
Yup, you'd be back to something like Hyper-V Replication (the simple way) or using a more complex tool like Starwind. Either way, you have a free storage clustering option.
Hyper-V replica is not clustering.
Sorry, wrong term. It's a "set".
-
@openit said in Hyper V replica VS Veeam B&R Replica.:
Okay. I will take help from Management to do maths about Redundancy requirement.
Let's assume, Server Redundancy is necessary after doing maths, could you please advice which one option for Server Redundancy is better ?What kind of redundancy do you need? RAID gives you redundancy, hence the "R".
What does management "NEED" according to their numbers? I'm sure they want the 5 nines of uptime, who doesn't... but what do the numbers say?
-
@Tim_G said in Hyper V replica VS Veeam B&R Replica.:
What does management "NEED" according to their numbers? I'm sure they want the 5 nines of uptime, who doesn't... but what do the numbers say?
I want six!
-
Important to note as Tim points out, if all you want is five nines of uptime, you can achieve that pretty inexpensively with non-redundant servers. One really good, well treated server with good RAID, good environment, and good management can average five nines.
-
@scottalanmiller said in Hyper V replica VS Veeam B&R Replica.:
Important to note as Tim points out, if all you want is five nines of uptime, you can achieve that pretty inexpensively with non-redundant servers. One really good, well treated server with good RAID, good environment, and good management can average five nines.
Five nines is just over 5 minutes downtime over a year. You can't update a server in under 5 minutes. You would need a cluster to move it to, or another server to take over in some way so you can down the original. But yeah you are right... I've seen single servers that are well treated environmentally and everything stay running for years. As a consequence they we never updated, but that's besides the point. Stability-wise, definitely way over five nines is possible from a single server as you said.
-
I have have effectively zero downtime in the past 10 years in my shop from server failures. I don't have any type of HA or fail over setup or vmotion type setup. We are not 24/7 shop, so taking the system down after hours to do maintenance doesn't count against our five 9's of uptime.
This is all in a onsite DC with UPS power backup and it's own AC unit. Definitely not as 'good' as a Level 3/4 DC, but not a hall closet either.
Modern servers, (starting at least 10, but probably more like 20 years ago) generally do pretty well stability wise as long as you aren't booting them all the time, and the temp/power remain mostly consistent.
-
@Dashrender said in Hyper V replica VS Veeam B&R Replica.:
We are not 24/7 shop, so taking the system down after hours to do maintenance doesn't count against our five 9's of uptime.
That's a good point.
When five nines really matter is during production hours... when end users and the business will be negatively impacted. If this is a web hosting company, then five nines matter 24/7. For a typical SMB, fine nines may only mean all services are 100% uptime for only half the day, leaving you with the other half for maintenance and other things.
-
@Tim_G said in Hyper V replica VS Veeam B&R Replica.:
@scottalanmiller said in Hyper V replica VS Veeam B&R Replica.:
Important to note as Tim points out, if all you want is five nines of uptime, you can achieve that pretty inexpensively with non-redundant servers. One really good, well treated server with good RAID, good environment, and good management can average five nines.
Five nines is just over 5 minutes downtime over a year. You can't update a server in under 5 minutes. You would need a cluster to move it to, or another server to take over in some way so you can down the original. But yeah you are right... I've seen single servers that are well treated environmentally and everything stay running for years. As a consequence they we never updated, but that's besides the point. Stability-wise, definitely way over five nines is possible from a single server as you said.
Your nines normally refers to your unplanned downtime, not the planned downtime. It's true, companies that run 24x7 and can never do a standard reboot will not always be able to do that. But normal companies have support greenzones (including Wall St banks) that they don't measure against downtime. Since things like rebooting or patching don't actually imply that the server is down, it's not counted.
Part of the importance here is defining if you mean "server downtime" or "application downtime" as they are importantly different at this level. A server rebooting is still up, but the application running on it is down.
-
@Tim_G said in Hyper V replica VS Veeam B&R Replica.:
@Dashrender said in Hyper V replica VS Veeam B&R Replica.:
We are not 24/7 shop, so taking the system down after hours to do maintenance doesn't count against our five 9's of uptime.
That's a good point.
When five nines really matter is during production hours... when end users and the business will be negatively impacted. If this is a web hosting company, then five nines matter 24/7. For a typical SMB, fine nines may only mean all services are 100% uptime for only half the day, leaving you with the other half for maintenance and other things.
Pretty common for it to not be over ten hours a day and often only six days a week. Most SMBs have at least one entire day of "free time."
Even FX trading on Wall St. has a GZ from ~7PM Friday night to 5AM Sunday morning. FX is the most demanding of all financial applications.
-
@Dashrender said in Hyper V replica VS Veeam B&R Replica.:
@openit said in Hyper V replica VS Veeam B&R Replica.:
@JaredBusch said in Hyper V replica VS Veeam B&R Replica.:
Honestly, IMO, from what little we know of the OP's environment, he does not need replication either. Just a single server and a backup.
I see.
Actually I have inquired with our management people about the "acceptable downtime for server", they said "one day" is okay. Here according to management (aka user) is meant for File Server and as you know, they are not aware of what DC, DNS etc. are.
Most of our production work depends on File Server, and based on above info. acceptable downtime for File Server will one day.
And for any hardware failure, the repair service will be next working day. The vendor from whom we have warranty tie up are working 5 days a week and we are working 6 days a week. If any failure happens at last working day on the week and spare part is not available immediately with them, we may consider around "3 days downtime for server to come up"
And you know, how the situation of IT guyz in this process.
"So I am thinking of Server Redundancy, for company benefit" and of course "to have piece of mind for myself
"Do you need the server itself to be up though? In your situation if I was going to have extended outage, I'd grab a PC with enough storage and install Hyper-v and then restore my data to that. Or look at a better warranty, like 4/6 hour response. That would be less expensive than a whole other server and one less box to worry about, that much less power usage, that much less worry about licensing, etc.
This one is also nice idea
I will keep it in my side note.So I need to have a PC with sufficient space. In emergency or required I will setup Hyper-v or restore the baremetal OS directly to physical pc and manage until Primary server is recovered. So here, in this case I need to make sure, I will copy back the whole data from Temp server (as it got updated with some data). Am I correct ?
-
@Dashrender said in Hyper V replica VS Veeam B&R Replica.:
Assuming your current server license is 2008 or newer, you get two VMs on one host.
With your current setup it sounds like you are using and services on a single install of Windows server, so you can move that to a VM, then create a second VM on that same host, install Veeam to it. I'd purchase a 2/4 drive NAS for the backup target.
You could do this all on your current server assuming is has enough resources (CPU, RAM, storage). Then purchase Veeam essentials ($850ish), NAS and drives ($1000ish).
It's Windows Server 2012 R2 standard. Yeah, we can it for two guests on Hyper-V.
And we do have NAS box for Backup target.
-
@Tim_G said in Hyper V replica VS Veeam B&R Replica.:
@openit said in Hyper V replica VS Veeam B&R Replica.:
Okay. I will take help from Management to do maths about Redundancy requirement.
Let's assume, Server Redundancy is necessary after doing maths, could you please advice which one option for Server Redundancy is better ?What kind of redundancy do you need? RAID gives you redundancy, hence the "R".
What does management "NEED" according to their numbers? I'm sure they want the 5 nines of uptime, who doesn't... but what do the numbers say?
If it's about RAID. We have RAID 1 on this Windows Server. I know RAID 10 is best option, but this is what we have currently. But still it is 1 Drive fault tolerant.
Well, this RAID is upto drive failure, it's not going to cover RAM/motherboard/processor etc failure, right ? that's why I was talking about whole separate server.
But as @Dashrender mentioned, a temp PC can do the trick in emergency.
-
@openit how long does it take you to get a new server to replace that one should it totally die (like go up in flames) or how long to get it fixed if only part of it dies?
-
@scottalanmiller said in Hyper V replica VS Veeam B&R Replica.:
@openit how long does it take you to get a new server to replace that one should it totally die (like go up in flames) or how long to get it fixed if only part of it dies?
If it's only about parts, it will depends on our Service Provider, from whom we have server warranty contract. According to them (service provider) , next business day response and I have no idea if they can fix it completely by next business day itself......
If it's about getting new server if main server totally dies, I am afraid it can take longer than repairing things. As I need to get quote, propose to management, approval, delivery and restoring the OS and Data from Backup. I believe management will not take long to respond on such critical or emergency cases.
-
@openit said in Hyper V replica VS Veeam B&R Replica.:
@scottalanmiller said in Hyper V replica VS Veeam B&R Replica.:
@openit how long does it take you to get a new server to replace that one should it totally die (like go up in flames) or how long to get it fixed if only part of it dies?
If it's only about parts, it will depends on our Service Provider, from whom we have server warranty contract. According to them (service provider) , next business day response and I have no idea if they can fix it completely by next business day itself......
If it's about getting new server if main server totally dies, I am afraid it can take longer than repairing things. As I need to get quote, propose to management, approval, delivery and restoring the OS and Data from Backup. I believe management will not take long to respond on such critical or emergency cases.
Well, a full server is what the vendor does when they can't fix it in their time frame, in theory.
-
@scottalanmiller said in Hyper V replica VS Veeam B&R Replica.:
@openit said in Hyper V replica VS Veeam B&R Replica.:
@scottalanmiller said in Hyper V replica VS Veeam B&R Replica.:
@openit how long does it take you to get a new server to replace that one should it totally die (like go up in flames) or how long to get it fixed if only part of it dies?
If it's only about parts, it will depends on our Service Provider, from whom we have server warranty contract. According to them (service provider) , next business day response and I have no idea if they can fix it completely by next business day itself......
If it's about getting new server if main server totally dies, I am afraid it can take longer than repairing things. As I need to get quote, propose to management, approval, delivery and restoring the OS and Data from Backup. I believe management will not take long to respond on such critical or emergency cases.
Well, a full server is what the vendor does when they can't fix it in their time frame, in theory.
huh - that's not my experience. I'll qualify this.
IBM server 13 years ago. Client ordered brand new server, I installed it, and it had problems (sadly now I can't recall the actual problem). IBM came out and replaced the RAM day 1, called on day 2 still have problems, they replaced the mobo, several hours later, confirmed not fixed, replaced RAID card on Day 3.
In the end we did get a new server after every part was changed except the chassis. And even then, I'm not sure it was IBM who replaced it, or the ITSP I worked for at the time.
-
@Dashrender said in Hyper V replica VS Veeam B&R Replica.:
@scottalanmiller said in Hyper V replica VS Veeam B&R Replica.:
@openit said in Hyper V replica VS Veeam B&R Replica.:
@scottalanmiller said in Hyper V replica VS Veeam B&R Replica.:
@openit how long does it take you to get a new server to replace that one should it totally die (like go up in flames) or how long to get it fixed if only part of it dies?
If it's only about parts, it will depends on our Service Provider, from whom we have server warranty contract. According to them (service provider) , next business day response and I have no idea if they can fix it completely by next business day itself......
If it's about getting new server if main server totally dies, I am afraid it can take longer than repairing things. As I need to get quote, propose to management, approval, delivery and restoring the OS and Data from Backup. I believe management will not take long to respond on such critical or emergency cases.
Well, a full server is what the vendor does when they can't fix it in their time frame, in theory.
huh - that's not my experience. I'll qualify this.
IBM server 13 years ago. Client ordered brand new server, I installed it, and it had problems (sadly now I can't recall the actual problem). IBM came out and replaced the RAM day 1, called on day 2 still have problems, they replaced the mobo, several hours later, confirmed not fixed, replaced RAID card on Day 3.
In the end we did get a new server after every part was changed except the chassis. And even then, I'm not sure it was IBM who replaced it, or the ITSP I worked for at the time.
Okay, it's what GOOD server vendors do when they can't figure it out. IBM couldn't get their own stuff working internally, so that doesn't surprise me. That's exactly why IBM doesn't depend on IBM gear themselves.
