 |
Token Ring to Ethernet LAN Migration If you are a token ring user, there are many reasons why you should consider migrating from token-ring to ethernet. First and foremost, ethernet LAN's are less expensive to build and operate. Ethernet PC network cards are cheaper than token-ring network cards and today's ethernet switches cost less per port than their token-ring counterparts. Second, the new gigabit-ethernet switches can provide greater bandwidth for your LAN. Finally, there are many new networking products that simply do not support token ring. OK, so you have decided to migrate from token ring to ethernet. Unless you have an unlimited budget, you will have to complete the migration in stages. This means that for a while you will be supporting both a token-ring LAN segment and an ethernet LAN segment. There are several technical issues associated with this. These include layer-two issues such as frame size, MAC addressing, and source-routing. There are also some layer-three IP routing issues. While these issues are not insurmountable, they can cause problems if you do not plan for them. The first layer-two issue with token ring and ethernet concerns their different data frame sizes. Token-ring data frames can be 4096 bytes in length, or longer. In contrast, ethernet frames are 1500 bytes or less. If a router is used to connect the two LAN segments, frame size is generally not an issue. If the router is routing IP, it will automatically segment large token-ring IP frames into smaller frames before routing them to the ethernet segment. The second layer-two issue is addressing. Ethernet adapters, like token-ring adapters, are shipped with "burned-in" Media Access Control (MAC) addresses. These addresses are six bytes in length, and are (at least, in theory) unique. Ethernet MAC addresses are LSB (Least Significant Bit) first while token ring MAC addresses are MSB (Most Significant Bit) first. While routers handle the conversion for the data frames they forward, you should be aware of the difference in bit order. The third layer-two issue is that token ring has an architectural feature not present in ethernet. This feature is known as source-route bridging. The token-ring data frame has a MAC-layer field called the Route Information Field (RIF). The RIF is used to store information about ring numbers and bridge numbers that the data frame has traversed to arrive at its destination. Ethernet data frames do not have this field. Layer-three issues related to a migration from token ring to ethernet can also be complex. During the migration, if a router is used to connect the ethernet segment to the token-ring segment, two IP subnets will be created. If you are also operating other routers on the LAN for a WAN connection or for Internet connectivity, this will result in the added complexity of multiple gateways. Aditional routing issues arise if you decide to "multihome" some of your key host systems and servers, so that they are directly connected to both ethernet and token-ring LAN segments. While this works well for some types of servers (Unix, AS/400, S/390), there are often issues with Microsoft servers. In short, if you are running a token-ring LAN, now may be the time to migrate to ethernet. However, you should give some thought to your coexistence strategy during the migration process. FLG Networking Services has completed several token-ring to ethernet migrations for clients and can help you through these technical issues. |
|
|
| ||
|
To discuss your networking requirements, email us at waltermorris3@yahoo.com OR call (408) 842-8630. |
||
