Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Riverstone Networks (RSTN) -- Ignore unavailable to you. Want to Upgrade?

To: ahhaha who wrote (81)	7/22/2001 7:38:47 PM
From: ahhaha	Read Replies (1) \| Respond to of 290

ENHANCEMENTS TO ROUTING In this section we discuss the enhancements made in support of GMPLS. These enhancements have been made to address some of the challenges of using MPLS to control optical and SONET/SDH time-division multiplexing (TDM) networks. These include: 1)The MPLS label space is comparatively large (one million per port), whereas there are a relatively limited number of lambdas and TDM channels (tens to hundreds per port today, scaling to thousands over the next few years). 2)MPLS LSPs can be allocated bandwidth from a continuous spectrum, whereas optical/ TDM bandwidth allocation is from a small discrete set of values. 3)Today there are rarely more than 10 parallel links between a pair of nodes. To handle the growth of traffic providers will need to deploy hundreds of parallel fibers, each carrying hundreds of lambdas between a pair of network elements. This in turn raises three sub-issues: a.The overall number of links in an optical/ TDM network can be several orders of magnitude larger than that of an MPLS network. b.Assigning IP addresses to each link in an MPLS network is not particularly onerous; assigning IP addresses to each fiber, lambda, and TDM channel is a serious concern, because of both the scarcity of IP addresses and the management burden. c.Identifying which port on a network element is connected to which port on a neigh-boring network element is also a major management burden and highly error-prone. 4)Fast fault detection and isolation, and fast failover to an alternate channel are needed. 5)The user data carried in the optical domain is transparently switched to increase the efficiency of the network. This necessitates transmitting control plane information decoupled from user data. Note that all of the above are issues for MPLS networks as well; however, these issues are immediate and pressing for optical networks. The following sections describe how GMPLS addresses these issues. To help understand the routing enhancements described below, we start with a brief description of link state protocols such as IS-IS and OSPF. Consider a network as a directed graph whose nodes are network elements (MPLS switches, cross-connects, etc.) and whose edges are links (fibers, cables, etc.). Each edge in the graph has associated attributes such as IP addresses, cost, and unreserved bandwidth. A link state protocol allows all the nodes to dynamically coordinate a coherent up-to-date picture of this graph, including the attributes of each edge. This picture of the graph is referred to as the link state database. Once the link state database is synchronized among all participating routers, each router uses the database to construct its own forwarding table. When a packet arrives at a router, the for-warding table is then consulted to determine how to forward the packet. Should the status of any link be changed, including adding or removing links, the link state database must be resynchronized, and all of the routers must recalculate their forwarding tables using the updated information in the link state database.