You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 11 Next »

2 high-performance computers, running Windows server.

ChemIT actions

1) Determine if there is rack space, and where it is.

2) Worth doing Hyper-V? (Oliver guesses probably not.)

Selection and pricing

Computer pricing considerations.

Note: No charge for Windows OS, due to Cornell's campus agreement.

Plus:

~$170 for a UPS (see below).

~$150 for network cables, including maybe a switch.

  • Racks are ~$550, but generally bought by those with bigger systems and not "sold back" for recovery.

Support and Risk management considerations

Remember these are two distinct servers, both running Windows Server.

OS maintenance windows

Proposed Download automatically. Install manually (by ChemIT or researcher?) so no surprised interruptions.

  • If ChemIT to do the updates, determine monthly schedule and stick to it so predictable to researchers.
    • If ~4hour window of interruption, should be fine if date/ time known well in advance to researchers. For both systems.

Security

Network configured to prevent off-campus access, unless using VPN (10-space).

  • No web, email, etc. access from the systems. (Not a "play" station!)
  • Systems still able to get updates via the network.

Leave Symantec Anti-Virus (SAV) off, for increased performance.

  • Researchers must practice safe computing practices, of course.

Robustness

Data

Confirm: Data not stored persistently on either of these computational systems, thus reducing the value in investing in these data options.

Data options

1) EZ-Backup: off-site, robust backup

  • EZ-Backup: $5.50/ month, for first 18 GB, compressed.

Reasons to not do this, other than cost:

  • Periodic performance hit, during daily backup window.
  • ChemIT will restore OS and applications without relying on backups since faster. So do //only// for researchers' data.
    • If all calculations are pulled off system right away, need for backups is thus much reduced.

2) 2 drives, mirrored, in each server. Via hard ware RAID, ideally.

  • If OK redoing last run if the single drive fails, no need to invest.
  • Can always buy a second drive later, if want to start doing this. (Only 2 drives per computer, see (3), below.)

3) File versioning, local. Would need another drive (need? want?).

  • Can get via saves to off-site file share, such as Petersen's SFS file share.
    • Size and quantity of data expected?
  • Can always buy a second drive later, if want to start doing this. (Only 2 drives per computer, see (2), above.)

UPS

~$170. Budget to replace every ~3 years or so.

GOAL: Protect against power outages measured in at least the single-digit minutes.

  • Both servers covered by one UPS.
    • Q: For how much time, versus one server per one UPS?
  • Only one server can be governed by UPS status.
    • Q: Invest in figuring out how to shut down second server elegantly, from UPS-governed server?

Access and permissions

Up to 2 Remote Desktop accounts at once, per server (plus console access).

  • Above that, must add and invest server resources to run Terminal Services.

Group's file management via NetID directories? Or simply move to group's file server?

Who all needs access, and to what?

  • No labels