| |
 |
|
 |
 |
|
- Can the server software operate on Unix e.g. Solaris or BSD flavours? Will we still receive the same level of support under Unix?
We have installed and tested Collector / Seeker on Windows 2000, Debian Linux, FreeBSD Unix, and Windows XP. Those are all the OSs we can be sure we support (as far as replicating any problems) right now for SenseMaker Collector. SenseMaker Explorer / Modeler have been tested on Debian Linux, Intel Mac, Windows 2000 and Windows XP. We can't yet guarantee replication of any problems on Sun Solaris.
- In your technical requirements document you say that RedHat is recommended for the server. Does this mean that when installed on Windows Server 2003 there is a performance loss, functionality loss or support issue?
There should be no losses of performance or functionality under Windows Server 2003, but there could be a loss of support as we don't have it running on our servers. We do think web-facing servers are better off in Unix or Linux than Windows, mainly for security reasons, and that is why we recommend not using Windows for Collector. But it should run the same there.
- Do you have a detailed installation guide for installing Windows Server 2003? Perhaps that includes the correct versions and configurations for Apache, Phython, Django and Python?
This is addressed in the detailed installation guide for SenseMaker Collector on Windows.
- Do you have any specific concerns of Unix security versus Windows in relation to your product?
No, nothing specific.
- As for capacity planning: I presume your application can use as much disk space and CPU cycles as required by the researchers. Any guidelines on capacity planning based on business need?
At our external hosting service we have a dedicated machine with a 2GHz CPU, 1GB RAM, 120 GB hard drive, and also a "shadow drive" which maintains a constant backup copy. We would suggest something similar as a *minimum* configuration. Any improvements in any of these specs would be useful for larger projects. If surveys are very large (ie, with many questions) and/or the number of people is large or excessively concurrent (ie, >1000) it is probably better to do more than the minimum.
- Will the recommendation of minimum server (2GHz CPU, 1GB RAM, 120 GB) will be adequate for average usage?
Collector's biggest strain so far was a project which collected 2600 stories with 40 indexes per entry (making over 100K data elements). Many of the stories were collected in a room with 40-60 people all entering data at the same time. The server handled this, however certainly a faster CPU would have been better.
- If the SenseMaker application is on a shared server with other web applications is there a potential to harshly impact these other applications? Do you suggest this should be on its own server?
We have never tested Collector running at the same time as other web applications. It runs alone on our server so we cannot guarantee results with any other configuration. Our best guess is that if you are running one survey at a time it should be fine to have other things running.
- Does SenseMaker tend to be I/O heavy or CPU heavy?
More CPU heavy. Also, when people are *exporting* data it is especially CPU heavy (so we recommend doing that when people are not using the system, like at night).
|
|
|
|
|
|
|
|
