We are trying to implement a remote backup copy solution that offers GFS and 3-2-1. Could someone explain to me please how backup copies work with incremental backups?
Our situation:
We have 11 Veeam backup jobs with around 50 VM's across them. Each job chains to the next so only one job has a scheduled start. Our existing system uses reversed incremental backups to USB 3 drives. We use 11 drives, 4 to cover each day for Mon-Thur, 4 to cover 1 per week each friday and 3 to cover the last day for each month in a quarter. We store the 2 more recent USB drives off site in a safe (no backup copies or replication).
Now we have a remote site with a 100Mbps (up and down) WAN link. We have a physical Veeam server at the remote site configured as a Veeam repository/proxy along with 5 hosts accessing a shared SAN. Our goal is to remove the need for USB disks by using Backup Copy jobs to copy Veeam files from the existing Veeam server to the remote site server.
My suggested design is to have 11 backup copy jobs, each job copying the data that matches its respective backup job. Each backup copy job would keep 4 restore points (to cover Mon-Thur) as well as 4 weekly backup restore points and 3 monthly backup restore points.
Initial tests are successful but potentially showing that we cannot copy the 4Tb's worth of Veeam job data to the remote site during the 18 hours each day available to run Veeam backup copy jobs (the other 6 hours are reserved for the Veeam server to run backups). So we are considering changing our Veeam backup process to use incrementals rather than reverse incrementals in the hope it would mean less data being copied over the WAN link.
I've read that incrementals are faster for backups than reversed incrementals and make smaller files, would the backup copy then just copy the incremental file?
In testing it seems that backup copy jobs cannot be chained, once created they remain "active" until they reach the synchronization interval. As a result we had in testing all 11 jobs running at once which caused the backup copy jobs to run very slowly. So I divided up the 18 hour window into time slots for each job. However then the majority of backup copy jobs failed because they didn't finish within the allocated time window. Via a trial I am about to use a WAN accelerator for the first time in the hope it will get the backup copy data job from the main site to the remote site quicker. I hope this will yield better results however I feel I'm still missing a trick in achieving what we want, hence my asking about using incremental backups rather than reversed incrementals, if I can get the original back ups completed quicker then I have more time to complete the backup copy data. Is there anything I should be doing differently?
We cannot be the only company looking to achieve this and I would welcome all advice, suggestions and feedback, hopefully all I have said makes sense
Could someone please give me some guidance? Thanks for reading