For the Boot Drive: If you want to boot your operating system from the drive you are about to partition, you will need:
A primary partition
One or more swap partitions
Zero or more primary/logical partitions
One or more primary/logical partitions
Zero or more swap partitions
Your boot partition ought to be a primary partition, not a logical partition. This will ease recovery in case of disaster, but it is not technically necessary. It must be of type 0x83 "Linux native". If you are using a version of lilo before 21-3 (ie, from the 1990s), your boot partition must be contained within the first 1024 cylinders of the drive. (Typically, the boot partition need only contain the kernel image.)
If you have more than one boot partition (from other OSs, for example,) keep them all in the first 1024 cylinders (All DOS partitions must be within the first 1024). If you are using a modern version of lilo, or a means other than lilo to load your kernel (for example, a boot disk or the LOADLIN.EXE MS-DOS based Linux loader), the partition can be anywhere. See the Large-disk HOWTO for details.
Unless you swap to files (see Section 9.2) you will need a dedicated swap partition. It must be of type 0x82 "Linux swap". It may be positioned anywhere on the disk (but see Section 4.4.3). Either a primary or logical partition can be used for swap. More than one swap partition can exist on a drive. 8 total (across drives) are permitted. See notes on swap size below (Section 4.4).
A single primary partition must be used as a container (extended partition) for the logical partitions. The extended partition can go anywhere on the disk. The logical partitions must be contiguous, but needn't fill the extended partition.
Everything in your linux file system can go in the same (single) partition. However, there are circumstances when you may want to restrict the growth of certain file systems. For example, if your mail spool was in the same partition as your root fs and it filled the remaining space in the partition, your computer would basically hang.
This fs contains spool directories such as those for mail and printing. In addition, it contains the error log directory. If your machine is a server and develops a chronic error, those msgs can fill the partition. Server computers ought to have /var in a different partition than /.
This is where most executable binaries go. In addition, the kernel source tree goes here, and much documentation.
Some programs write temporary data files here. Usually, they are quite small. However, if you run computationally intensive jobs, like science or engineering applications, hundreds of megabytes could be required for brief periods of time. In this case, keep /tmp in a different partition than /.
This is where users home directories go. If you do not impose quotas on your users, this ought to be in its own partition.
This is where your kernel images go. See discussion above for placement on old systems.
With ext2, partitioning decisions should be governed by backup considerations and to avoid external fragmentation Section 10.4 from different file lifetimes.
Files have lifetimes. After a file has been created, it will remain some time on the system and then be removed. File lifetime varies greatly throughout the system and is partly dependent on the pathname of the file. For example, files in /bin, /sbin, /usr/sbin, /usr/bin and similar directories are likely to have a very long lifetime: many months and above. Files in /home are likely to have a medium lifetime: several weeks or so. File in /var are usually short lived: Almost no file in /var/spool/news will remain longer than a few days, files in /var/spool/lpd measure their lifetime in minutes or less.
For backup it is useful if the amount of daily backup is smaller than the capacity of a single backup medium. A daily backup can be a complete backup or an incremental backup.
You can decide to keep your partition sizes small enough that they fit completely onto one backup medium (choose daily full backups). In any case a partition should be small enough that its daily delta (all modified files) fits onto one backup medium (choose incremental backup and expect to change backup media for the weekly/monthly full dump - no unattended operation possible).
Your backup strategy depends on that decision.
When planning and buying disk space, remember to set aside a sufficient amount of money for backup! Unbackuped data is worthless! Data reproduction costs are much higher than backup costs for virtually everyone!
For performance it is useful to keep files of different lifetimes on different partitions. This way the short lived files on the news partition may be fragmented very heavily. This has no impact on the performance of the / or /home partition.
Conventional wisdom creates swap space equal to the amount of RAM.
But keep in mind that this is just a rule of thumb. It is easily possible to create scenarios where programs have extremely large or extremely small working sets (see Section 3.5). For example, a simulation program with a large data set that is accessed in a very random fashion would have almost no noticeable locality of reference in its data segment, so its working set would be quite large.
On the other hand, a graphics program with many simultaneously opened JPEGs, all but one iconified, would have a very large data segment. But image transformations are all done on one single image, most of the memory occupied by the program is not accessed. The same is true for an editor with many editor windows where only one window is being modified at a time. These programs have - if they are designed properly - a very high locality of reference and large parts of them can be kept swapped out without too severe performance impact. A user who never never quits programs once launched would want a lot of swap space for the same reason.
Servers typically are configured with more swap space than their desktop counterparts. Even though a given amount of swap is sufficient for its operations, the server might come under transient heavy loads which cause it to page out at a high rate. Some administrators prefer this to the server crashing altogether. In these cases, swap might be several times the size of ram.
Currently, the maximum size of a swap partition is architecture-dependent. For i386, m68k, ARM and PowerPC, it is "officially" 2Gb. It is 128Gb on alpha, 1Gb on sparc, and 3Tb on sparc64. An opteron on the 2.6 kernel can write to a 16 Tb swap partition. For linux kernels 2.1 and earlier, the limit is 128Mb. The partition may be larger than 128 MB, but excess space is never used. If you want more than 128 MB of swap for a 2.1 and earlier kernel, you have to create multiple swap partitions (8 max). After 2.4, 32 swap areas are "officially" possible. See setting up swap for details.
footnote: "official" max swap size: With kernel 2.4, the limit is 64 swap spaces at a maximum of 64Gb each, although this is not reflected in the man page for mkswap. With the 64 bit opteron on the 2.6 kernel, 128 swap areas are permitted, each a whopping 16 Tb! (thanks to Peter Chubb for the calculation)
The short answer is anywhere is fine. However, if you are interested in extracting as much speed as possible, there are two basic strategies (other than buying more RAM).
Split the swap space across multiple drives, or at least on the drive you write to least.
Put each swap partition on the outer tracks.
Here are the considerations:
If you have a disk with many heads and one with less heads and both are identical in other parameters, the disk with many heads will be faster. Reading data from different heads is fast, since it is purely electronic. Reading data from different tracks is slow, since it involves physically moving the head.
It follows then that writing swap on a separate drive will be faster than moving the head back and forth on a single drive.
Placement: Older disks have the same number of sectors on all tracks. With these disks it will be fastest to put your swap in the middle of the disks, assuming that your disk head will move from a random track towards the swap area.
Newer disks use ZBR (zone bit recording). They have more sectors on the outer tracks. With a constant number of rpms, this yields a far greater performance on the outer tracks than on the inner ones. Put your swap on the fast tracks. (In general, low-numbered cylinders are associated low partition numbers. However, see Kristian's more recent comments on this issue. -Tony)
Usage: Of course your disk head will not move randomly. If you have swap space in the middle of a disk between a constantly busy home partition and an almost unused archive partition, you would be better of if your swap were near the home partition for even shorter head movements. You would be even better off, if you had your swap on another otherwise unused disk, though.
Striping: Speed can be increased by writing to multiple swap areas simultaneously. Swap spaces with the same priority will be written to like a RAID. See Section 9.3.
Summary: Put your swap on a fast disk with many heads that is not busy doing other things. If you have multiple disks: Split swap and scatter it over all your disks or even different controllers.