I forgot to add, the 1000+ user sites have either 1G or 10Gig connectivity. Up until last month, I was at an agency, without a full view of the State's architecture.
-----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Crown, David T. (DTI) Sent: Monday, July 1, 2013 16:08 To: [email protected] Subject: RE: [mssms] Sizing Question Trevor, I was planning on using the default policy refresh, and daily hw inventory. The lower performing storage is still a fibre channel box, just not a monster box that I'm putting SQL on. I can know I can get 5 to 10K IOPS, with a ceiling of 20K for bursts. -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Trevor Sullivan Sent: Monday, July 1, 2013 11:52 To: [email protected] Subject: RE: [mssms] Sizing Question David, One thing I'd point out: Don't use "lower performing storage" for the Content Library, and your package source. This can get hit pretty hard with large package distributions (eg. WIM files, Office, et al.). You should also consider this question: Are you planning on using default schedules for policy refreshes and inventory? This will impact your sizing for management points. As for your remote sites, consider avoiding the usage of Secondary Sites, as long as there is a large enough WAN pipe. If you assume 1,000 users, have 100Mbit pipes (or larger), and don't have aggressive policy refresh and inventory schedules, then you should be alright with just a Distribution Point. After all, the vast majority of your network utilization will be package pull requests. Cheers, Trevor Sullivan -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Crown, David T. (DTI) Sent: Monday, July 1, 2013 8:33 AM To: '[email protected]' Subject: [mssms] Sizing Question I know it being Monday and all, I'm wondering if you find folks wouldn't mind double checking my sizing. I changed roles with my employer, and the scale of the configmgr environment I'm used to supporting has grown quite a bit. I'm looking to migrate about 15K, scaling to a potential of 40K, clients from a 2007 and a 2012 (with a CAS for political reasons) to a single 2012 site. The plan is to use a box with 16 cores and 48GB (the plan is get it to 96) of ram with SQL on box and no other roles. For the backing disks, I was looking at three Fibre Channel Luns. One 300~500 gig disk on a high performance lun that can sustain 20K IOPS for the site directory and sql, one 2TB lun on some lower preforming storage for package source and the content library, and a third lun in the same lower performance tier (~1TB) for my backups. I plan on using two to four MP's as vm's behind a load balancer, two unprotected DP's for failback, and one SUP. As the environment is deployed, I plan on bringing in protected DP's, and for my larger sites (500 to 1000 users), I plan on sticking a secondary at the site. So my question is does my proposed environment look like it could support the number of clients I'll be managing?
