Benches Deployment and MDM

2

Server hardware and CPU load

apple Verified

Imported from the Apple Support Forums via the Wayback Machine 🔗

Hi

I'm in the process of putting together a recommendation for a client to purchase Final Cut Server for their video production workflow, and I'm trying to determine the best hardware to install the server on.

Right now they have an XServe 2,1 which is hosting a web server with some transcoding features. It's not heavily taxed, but it is mission-critical and directly accessed by their clients' websites.

Their storage solution is a Terrablock SAN array, with fiber links to the XServe and three Mac Pros.

I don't have a lot of experience with fiber SANs, so one question I'm not sure about is: does the hardware running FCS have to have direct fiber connection to the storage, or is it enough that the client machines have a fiber connection? In other words, with such a system, must the server hardware be fiber-capable (ie, Mac Pro or XServe), or will, for example, the mini+SL Server solution work?

A connected question - what kind of load does FCS place on the server's CPU? If we do need to use the existing XServe (the budget may not stretch to a second one), I don't want to jeopardize the existing server tasks.

many thanks for any thoughts or advice -

Matt

2 replies

Sign in or join Beachball Help to reply.

  1. 0

    Apple Support Community member

    Imported from the Apple Support Forums via the Wayback Machine 🔗

    Taking your comment "I don't want to jeopardize the existing server tasks" to heart, the only solution is to invest in an additional Xserve.

    It is not required for Final Cut Server to be on Xsan, but it will create a bottleneck if it is not. Note also that Terrablock is a volume-locking SAN, which will be a little trickier to set up and manage from a workflow perspective.

    There are two main tasks for Final Cut Server - running the database (updating, searching), and transcoding video (generating proxies, converted outputs). RAM is affected by the size of the database and the number of simultaneous encodes. CPU is affected by number of users/searches on the database, and the complexity/length of the video encodes. Both are important in most situations.

    The Mac mini is inappropriate since it is not server class hardware - it has a slow CPU, insufficient RAM, a slow 2.5" drive, and an integrated graphics chip, which by itself makes it incompatible with the Final Cut Server requirements.

    On a related note, I'd also be a little concerned about their existing "mission critical" Web server - a single point of failure that is exposed to external networks is something they should probably address at some point...

  2. 0

    Apple Support Community member

    Imported from the Apple Support Forums via the Wayback Machine 🔗

    John F. Whitehead wrote:
    Taking your comment "I don't want to jeopardize the existing server tasks" to heart, the only solution is to invest in an additional Xserve.

    I think ultimately that may be the answer. For testing purposes, and to get the client started with metadata creation and getting used to the workflows, I will probably install on one of the client Mac Pros.

    It is not required for Final Cut Server to be on Xsan, but it will create a bottleneck if it is not.

    In this case, we're going to be using edit-in-place workflows, so the question I'm trying to get my head around is whether an ethernet connection between Server machine and SAN is sufficient for the database and proxy transcoding operations, or whether it NEEDS fiber. Possibly a question that can only be answered by testing at the client.

    Note also that Terrablock is a volume-locking SAN, which will be a little trickier to set up and manage from a workflow perspective.

    The new Facilis software implements a "shared file system" bringing file-level locking to the SAN, so this may be less of an issue. I'm still not 100% clear, though, as to which "user" is checking in/out the files, as far as the SAN is concerned. Is it the user on the client machine, or the user logged into the FCSvr +client software+?

    I'd also be a little concerned about their existing "mission critical" Web server - a single point of failure that is exposed to external networks is something they should probably address at some point...

    Yeah, definitely a conversation to be had...

    I appreciate your time and thoughts.

    Matt