GIS workstation setup tips

A page to collect best practice tips for setting up a GIS workstation. Page is under on-going construction

Hardware

CPU

Multi-core is in and there are many steps(grades or variants), if you plan to run virtual machines make sure to get something with the virtualization instruction set.

RAM

The more the better, but it does depend on what applications you plan to use. Things like R benefit more from RAM than some applications that are more graphic intensive.

Video Card

Quality, dual screens is a big draw for GIS analysts to see a map on one screen and a table on another. If you do mostly GIS number crunching via commandline you probably don't need a beefy card, unless you want to learn to program GPU's

Hard Drives

Plan for double what you actually think you need. Why? Backups and room to grow. SATA is by far the cost most effective solution which has a good balance of speed/price/size. 2TB will cost $1000 or less these days.

Consider using a smaller faster drive for your OS and larger slower drives for Read Only data storage.

Raid Drives

Raids are only good if:

1. It's hardware based
2. You run drive health monitoring tools on a regular basis AND
3. You know how to rebuild it if it fails

Operating System

Are there any advantages to choose a specific operating system (OS)?

I've been using Linux at home and at work for several years, so I thought I would share some thoughts here. In my own humble opinion Debian/Ubuntu is the best Linux Operating System for new Linux users. I've found the community at Debian (via the mailing list) to be one of the most supportive of all the online communities I've been a part of. There are plenty of people on that list willing to share their knowledge. Ubuntu does a better job of packaging the Debian operating system, and they make support for hardware that requires proprietary drivers much easier. I also find that the Ubuntu software package repositories often contain software I need, which is not always in the Debian repository.

When I first was introduced to Linux I tried a few other distributions. This included ArkLinux, Fedora, OpenSuse and Gentoo. After some advice from my local Linux User Group I tried Debian. I don't think I'd ever have been successful with Linux if it hadn't been for Debian.

A couple other things to note:

• What most people think of as "Microsoft Windows" is actually an operating system, desktop, and suite of office applications packaged together. In Linux these these are all different components. Linux is the operating system, while the desktop used is a separate component, like Gnome or KDE. You can also use different office software, like Open Office or AbiWord.

• All Linux distributions essentially build on the same operating system. The main distinction between them is the way they manage software installation, which is quite different then on Microsoft Windows. Linux distributions can be divided into two (2) main camps. One camp uses the Debian format for installing and managing software, while the other uses the Red Hat format. The other factors that distinguish an operating system are how often they relese new versions of the distrubution and how they handle "non-free" software. These two factors help to distinguish Ubuntu and Debian. Debian makes a new release whenever it is ready, not on a regular schedule, while Ubuntu makes regularly scheduled releases. Debian takes a somewhat "hostile" view to free software, while Ubuntu does not.

• One of the main challenges to using Linux is hardware compatibility. Before you buy any hardware or periphials (scanners, network interface cards, microphones, scanners, video cards) for a computer on which you plan to run Linux make sure it is compatible with Linux. Many companies make Linux specific drivers for their hardware, while others have hardware that doesn't play with Linux well. Almost all companies make hardware that plays with Microsoft's operating systems well. This isn't the case with Linux. Don't buy hardware for Linux and expect it to "just work" unless you have done your homework.

The Sunburned Surveyor

32 vs 64 bit

If you have or plan to have 4GB of RAM or greater 64 bit is necessary to correctly address the RAM and use the system to it's potential. For windows this might mean some slightly buggy drivers and the need to run some 32 bit applications in compatibility mode (does this work). For linux at this point you're pretty much taken care of. Occasionally some applications aren't compiled or take longer to reach 64 bit platforms (like Adobe Flash) but that trend is decreasing. Some compiling may be necessary to get what you want, but that is true for all systems in order to have the latest patches a tools.

Partitions

Tips at a glance

• One separate partition for the OS
• On GNU/Linux OSes an extra swap partition is required. Recommended size is equal to or double as much as the system's available RAM.
• At least one separate partition for data storage
• If a second (directly attached to the motherboard) or a fast external hard disk drive is available, then data can be stored on it (not the one that carries the OS) to protect system resources from being consumed 100% when the computer executes heavy GIS processing tasks which will slow down the system and prohibit the use of other programs.
• In case of multi-user access on projects, easy data exporting/sharing and can be achieved using for example the NFS network filesystem protocol
• Create at least 2 partitions / and /home to make upgrades and file management easier.

Details

Before even installing an operating system on a computer, the hard disk drive needs to be partitioned. For practical as well as for data safety/security reasons it is recommended to divide the hard disk's available space into (at least) two or more partitions. One partition to carry the operating system itself and another partition to store data. The size of the partition that will carry the operating system depends on the requirements of the operating system that is to be installed. The size of the data partition(s) depends on the user's needs.

• Which disk filesystem type?

There are several disk filesystem types which often are OS relevant/specific. More information can be found at wikipedia: http://en.wikipedia.org/wiki/File_system

More about partitions on GNU/Linux

• Which filesystem is better(=safer/faster)?

Among the most famous disk filesystems used on GNU/Linux OSes, are the ext3 and the XFS. The ext3 filesystem type seems to be faster than other types in several fields. On the other hand the XFS, although used mainly on servers, seems to perform the best overall highscore. [Sources: http://www.debian-administration.org/articles/388, http://linuxgazette.net/122/piszcz.html ]

- Note: As of Ubuntu 8.10 JFS is also supported, originally developed by IBM it's reported to be faster than ext3.

• What about Linux' swap?

In addition, on a GNU/Linux operating system one needs to define a swap partition. It is suggested to give as much space as the computer's available Random Access Memory (RAM) or, even better, the double. To exemplify, if a system features 1GB RAM then the swap partition should be sized between 1 and 2GB. [ More details at https://help.ubuntu.com/community/SwapFaq ]

- Note: This depends entirely on how you intend to use the machine. If you have a lot of ram (16GB) and want to always run everything in it, make the swap very small. Once you get past 2GB there is little or no reason to increase the swap(except if you use the suspend mode).

LVM

Logical_volume_management lets you arbitrarily span multiple drives of any size to make it look like one drive.

Pro:

• You can keep shoving disks in as you need and simply add on to the existing mount points

Con:

• You data can get spread across a lot of drives and if one goes bad in the middle it can be awkward to shuffle.
• A little more difficult to setup to start.

Backing up data

How often should one back up data?

How should one back up data?

• simple method: tar cjvf
• syncing method: using rsync

Offsite backup?

Other important issues

• Use a Universal Power Supply(UPS) and configure it to trigger safe shutdowns on extended power failure.

Note: A Good GIS system can be very similar to a good gaming system, poke around the internet for what people use for those.

Related links

• Discussion for creating this wiki-page at the GRASS-user mailing list:

http://n2.nabble.com/Tips-for-setting-up-an-new-FOSS-GEO-linux-box-tc1885845.html#none

• Some "historical" discussions upon the subject can be retrieved by searching within the GRASS-user mailing list for the keyword "workstation".