OSPF vs. ISIS

Long ago, when I didn’t know much about ISIS, and I mostly worked with customers running OSPF, I used to think of ISIS as a strange routing protocol, that nobody really cared much about. But I was sooo wrong!

I started studying ISIS in depth as I progressed in the service provider certification tracks, and obviously when I got to the E-level I had to know ISIS very well. Then, I had the chance to work for a major Service Provider in the US, and discovered that not many people actually knew much about OSPF, but were well versed in ISIS.

The more I learned about it, and the more I worked with it supporting my customer, the more I liked it. I came to realize that ISIS is actually a pretty awesome routing protocol. And I would like people to understand it better, and embrace it. I am sure that anyone who’s being in the Service Provider world, is already on board.

Students have always asked me: “If I am familiar with OSPF, would that help me understand ISIS? Do the concepts in OSPF, apply to ISIS? or are they way different?”

I guess I will answer those questions with YES! YES! and YES!

In some ways, the two protocols are very similar, and the concepts you already know in OSPF apply very closely, but at the same time, they are very different.

So, let me try to summarize the similarities and differences for you:

SIMILARITIES:

Both routing protocols are IGP (Internal Gateway Protocols) and Link State Protocols, which means they both advertise link information and build a Link State Database. The exchange of information is reliable, and the routers will make sure that the databases are synchronized.

They both run the SPF (Shortest Path First) algorithm to calculate the best path to each destination network, and to make that determination they add the metrics of the different links.

Also, before routers can exchange actual link information they form an adjacency, or formal neighbor relationship, which is built using hello packets. Both OSPF, and ISIS use hello packets for discovering neighbors and maintaining the relationship with those neighbors.

Also, when you implement OSPF or ISIS, you do so following a 2-level hierarchical model (backbone/non-backbone areas in OSPF, and LEVEL 1/LEVEL 2 in ISIS)

They also have many common features and options, just implemented in different ways.

To summarize, both protocols:

• Are IGP Link State routing protocols and build a Link State Database
• Have reliable updates
• Run SPF to calculate best routes.
• Use hello packets to create adjacencies
• Have a two level hierarchical model
• Support authentication and route summarization
• Use multicast addresses
• Elect a designated device on multiaccess networks.
• Can differentiate internal and external routes and tag routes
• Support features like graceful restart, GRES/NSR, Traffic Engineering, IPv6, and so on.

Thus, yes, there are a lot of similarities! But, as I said: they are also very different. Let’s take a look at that now:

DIFFERENCES:

ROUTER TYPES: * Internal router: A router with all its interfaces within the same area * Backbone router: A router with at least one interface connected to the backbone area. * Internal Backbone router: A router that with all its interface within the backbone area. * Internal Non-backbone router: A router with all its interfaces within an area other than the backbone * ABR (Area Border Router): A router connected to more than one area (commonly one of those areas is the backbone area) * ASBR (Autonomous System Boundary Router): Any router injecting routes into the OSPF domain via redistribution.	ROUTER TYPES: * L1 router: A routers that can only form adjacencies with other routers in the same area. Similar to an internal router. * L2 router: A router that can form adjacencies with routers in the same area & with router in other areas. Similar in concept to a internal backbone router. * L1/L2 router: A routers that can form adjacencies with L1 routers and also with L2 routes. Similar in concept to an ABR. NOTE: refer to the adjacency formation section for more details.
INTERFACE TYPES: * p2p * broadcast (LAN) – default for ethernet * NBMA * p2mp	INTERFACE TYPES: * p2p * broadcast (LAN) – default for ethernet
BROADCAST NETWORK:	BROADCAST NETWORK:
* A Designated (DR) and a Backup Designated Router (BDR) are elected	* A Designated Intermediate System (DIS) is elected.
* Routers establish adjacencies (Full State) only with the DR & BDR in the broadcast network * DR and BDR have adjacencies with all routers in the network	* Routers establish adjacencies with ALL other routers in the broadcast network.
* There is a BDR	* There is NO Backup (DIS)
* There is a DR/BDR per segment	* DIS per level, per segment
* Default priority for DR election = 128 (range 1-255) * Router with highest priority is elected. * If priority is the same, then router with highest router-id is elected * Priority of 0 means ineligible	* Default priority for DR election = 64 (range 0-127). * Router with highest priority is elected. * If priority is the same, then router with highest MAC address is elected. * Priority of 0 does NOT mean ineligible
* There is no preemption for the DR	* There is preemption for the DIS
PROTOCOLS SUPPORT: * V2 only supports IPv4. * V3 supports IPv4 and IPv6	PROTOCOLS SUPPORT: * IPv4, IPv6 & CNLS
ADJACENCY FORMATION REQUIREMENTS/RULES: * Same Area ID * Same timers (hello and dead intervals) * Same Area type (options bits) * Authentication type, and key * Unique RID * Matching MTU * Matching IP subnet and subnet mask	ADJACENCY FORMATION REQUIREMENTS/RULES: * For LEVEL 1 adjacencies: routers must be in the same area * For LEVEL 2 adjacencies: routers can be in the same area or in different areas. * By default, a Juniper router is configured as an L1/L2 router. => By default, between two Juniper router running ISIS, two adjacencies will be formed (L1 and L2) * Minimum MTU of 1492 is required * NET configured under lo0.0 must be configured In the example, all routers are L1/L2 but an L1 adjacency between R1 and R4 cannot form because they are in different areas.
AREAS:	AREAS:
* Area assignment: – An internal router has all interfaces within an area. – An ABR has interfaces in the backbone area, and interfaces on one or more non-backbone areas. – Boundary between areas is on the ABR.	* Area assignment: – A router is completely within an area. – All interfaces in the same area (always). – Boundary between areas is on the links.
* Area ID: – Associated with the interfaces – An interface belongs to only one are (a secondary area is possible with multiarea adjacencies)	* Area ID: – Associated with the router not the interfaces – A router can belong to multiple areas
* Inter-area information: – ABRs advertise inter-area routing information to internal routers using LSAs type 3 – If an area is configured as a stubby no-summary (totally stubby area), these LSAs are replaced with a default route, if configured NOTE: refer to the DEFAULT ROUTE section. – Policies can be applied to control which LSA3 are created by the ABR.	* Inter-level information: – L1/L2 routers do NOT advertise L2 routes to L1 routers by default (Behavior is equivalent to NSSA no summaries). > A policy can be configured to leak routes from LEVEL 2 into LEVEL 1 – L1/L2 routers advertise internal L1 routes to L2 neighbors by default. – L1/L2 routers do NOT advertise external L1 routes to L2 neighbors by default. > A policy can be configured to stop routes from LEVEL 1 into LEVEL 2 or to leak external routes. NOTE: with wide-metrics-only there is no difference between internal and external routes.
* Area Types: – Regular Area: LSAs type 1, 2,3,4, & 5 – Stub Area: LSAs type 1, 2, & 3 – Stub Area no-summary (totally stubby area): LSAs type 1, 2, & 3 (only 0/0) – Not So Stubby Area (NSSA): LSAs type 1, 2, 3, & 7 – NSSA no-summary: LSAs type 1, 2, 7, & 3 (only 0/0). 0/0 can be changed to 7.	N/A
HIERARCHICAL DESIGN:	HIERARCHICAL DESIGN:
* Area 0 required * Backbone – > Non Backbone area * Boundary = ABR	* Area 0 NOT required * LEVEL 2 -> LEVEL 1 * There is no backbone or area 0 configured * Boundary = L1/L2 router.
TIMERS:	TIMERS:
* Hello interval: – 10 sec (broadcast and p2p) – 30 sec (NBMA) Configurable	* Hello interval: – 3 seconds (for DIS routers) – 9 seconds (for non-DIS routers) Configurable
* Dead interval (3 x hello interval): – 40 sec (broadcast and p2p) – 120 sec (NBMA) Configurable	* Hold time (3 x hello interval): – 9 seconds (for DIS routers) – 27 seconds (for non-DIS routers) Configurable
* Age starts at 0 and increments up to maxage.	* Age is set to max. age and gets decremented up to 0
* Maximum Age = 3600 sec (60 min) Configurable	* LSP Lifetime = 1200 sec (20 min) Configurable
* Default LSA Refresh Interval = 3000 sec (50 min) Configurable	* Default LSP Refresh Interval= 883 sec (lifetime minus 317)
N/A	* CSNP Interval: – on broadcast interfaces = 10 sec – on point-to-point interface = 5 sec
ROUTE PREFERENCES: * Internal = 10 * External = 150 Configurable	ROUTE PREFERENCES: * Level 1 internal = 15 * Level 2 internal = 18 * Level 1 external = 160 * Level 2 external = 165 Configurable
OVERLOAD: indicated with metric = 65535	OVERLOAD: Indicated with overload bit
METRICS: * Not specified by standard * Referred as cost. * Most vendors use cost = 10^8/Bandwidth (by default), where 10^8 is the reference bandwidth. * Default metric for lo0 interface = 0 * You can configure the metric of the interface, the reference bandwidth, or the bandwidth of the interface.	METRICS: * Specified by standard ISO 10589 * 4 types of metrics defined: –Default:  Mandatory. Default for ALL interfaces = 10 (configurable value) – Delay: not used – Expense: not used – Error: not used * You can also configure the metric to be calculated automatically as 10^8/Bandwidth * Metrics can be configured on a per level basis.
* Maximum metric = 65,535	* Maximum Total Metric on a path = 1023 * Maximum metric on an interface = 63 Limit technically removed with wide-metric-only (~16 million maximum value)
DESTINATION ADDRESS/PROTOCOL: * Encapsulated within IP PACKETS using: – Protocol = 89 – Multicast DA = 224.0.0.6 (DR/BDRs) – Multicast DA = 224.0.0.5 (All OSPF routers)	DESTINATION ADDRESS/PROTOCOL: * Encapsulated within Layer 2 frames using: – DSAP/SSAP= 0xFE – Multicast DA = 0180.c200.0014 (LEVEL 1) – Multicast DA = 0180.c200.0015 (LEVEL 2) * Requires GRE encapsulation to run on connections such as IPSEC tunnels.
DEFAULT ROUTE: * When configured, an ABR can inject a default route into a stub area or a NSSA, – For Stub Area => LSA type 3 – For NSSA => LSA type 7 – For NSSA no-summary => LSA type 3	DEFAULT ROUTE: * An L1 router installs a default route in its routing table when it receives an update from an L1/L2 router with the attached bit set. * Can be disabled with the ignored-attached-bit command
ROUTER ID: * primary address of lo0.0 interface OR * router-id configured under routing-options.	ROUTER ID: * iso address (NET) of lo0.0 interface
TRAFFIC ENGINEERING: * Supported, but needs to be configured. * Use LSA Type 10.	TRAFFIC ENGINEERING: * Supported by default. * Use TE Sub-TLV within the Extended IS Reachability (TLV 22) and the Extended IP Reachability (TLV 135)
ROUTING POLICIES:	ROUTING POLICIES:
* Default Export Policy: – ALL routes are rejected = NO redistribution by default – An export policy would enable redistribution and create LSAs type 5/7 – LSAs are still created for interfaces running OSPF, and sent out (cannot modified by policies). – Writing and applying a policy that rejects everything: does NOT stop the advertisement of interfaces running OSPF	* Default Export Policy: -ALL routes (except direct routes for interfaces running ISIS) are rejected = NO redistribution by default – LSPs are created for interfaces running ISIS, and sent out (can be affected by policy). – An export policy can control which prefixes are actually advertised (even for interfaces running ISIS), configures redistribution and creates external routes – w/ narrow metrics) – Writing and applying a policy that rejects everything: stops the advertisement of interfaces running OSPF
* Default import policy: – All routes from SPF calculations are accepted and imported into the routing table. – LSAs sent by neighbors are accepted and placed in the LSDB, and SPF calculates routes (not affected by policies). – Import policies can be configured, but only for external routes.	* Default import policy: – All routes from SPF calculations are accepted and imported into the routing table. – LSPs sent by neighbors are accepted and placed in the LSDB, and SPF calculates routes, routes. (not affected by policies). – Import policies could NOT be configured for ISIS until Junos 17.1
AUTHENTICATION: * Simple test OR MD5 OR IPSEC * Per area OR per interface (all router in an area must have same authentication configured) * Applies to ALL packets	AUTHENTICATION: * Simple test OR MD5 * Per level or per interface: – Per level applies LSPs, CNPs and Hello messages (can be disable for a specific type of packet) – Per interface applies only for hello messages
ROUTE AGGREGATION: * area-range under the area configuration level Summarizes the prefixes injected into other areas by the ABR (LSAs type 3). ABR learns about these prefixes from LSAs type 1 and type 2. * area-range under the nssa area configuration level Summarizes the prefixes injected into an area, by the ABR, when translating from LSAs type 7 into LSAs type 5. * NO other LSAs can be summarized! * ONLY the ABR can summarize! * Specific prefixes within the range automatically suppressed. * The restrict option can be used with the area range to stop any updates for prefixes within the range.	ROUTE AGGREGATION: * NO specific command for route summarization. * To summarize prefixes 1) Create an aggregate and redistribute with policy 2) Redistribute the aggregate into ISIS using a policy:
ROUTING INFORMATION:	ROUTING INFORMATION:
* Packet Types: – Type 1 = Hello – Type 2 = Database Description – Type 3 = Link State Request – Type 4 = Link State Update – Type 5 = Link State Acknowledge	* Packet Types (PDU: Packet Data Units): – Hello broadcast network: For L1 and L2 – Hello point-to-point network – LSPs (Link State Packets): Like OSPF LSAs For L1 and L2 Include reachability TLVs. – PSNP (Partial Sequence Number PDU): Like OSPF Request/Update/Ack – CSNP (Complete Seq. Number PDU): Like OSPF DBD Only sent by DIS on broadcast net. Periodic For L1 and L2
* LSA Types: – Type 1 = Router LSA (all routers) – Type 2 = Network LSA (DR) – Type 3 = Network Summary LSA (ABR) – Type 4 = ASBR Summary LSA (ABR) – Type 5 = External LSA (ASBR) – Type 7 = NSSA LSA (ASBR NSSA)	* LSP Types: – Level 1 – Level 2
* Database on boundary router: An ABR has links state databases for all the areas it is connected to.	* Database on boundary router: A L1/L2 router has a database for L1 and a database for L2.
* SPF calculation on a per area basis	* SPF calculations on a per level basis