Active network

Service-enabled Networks From The Network to My Network

Four wheels traction CAGR to 2001 Internet Telephony Bring more quality Optical Internet Bring more capacity eBusiness Bring faster cycle times Wireless Internet Extend reach

What’s the scoop —the techies’ view TTM and product cycles make software reuse a must—enough stovepipes, layers are back with a vengeance! Very high-value traffic coexists with low value, low pay off traffic, all within the same commodity infrastructure Increasingly, “impedance mismatches” in the network generate opportunities for intelligent edge adaptation Standardization lags behind the mighty innovation curve— the IETF process hardly matches a trillion dollar business Network wires are now much faster than I/O busses—it’s time to bring the optical revolution into servers/storage

What’s the scoop —the biz view ASPs and eBiz test the market for new services, non stop— many fail, but they’ve thrown a monkey-wrench into the net Customers demand new soft features in network gear—manufacturers cannot keep up with fast (contrasting) reqs Personalized customer care stipulations are strategic for customer expansion and customer loyalty Hardware innovations are fueling a steep innovation curve— what a great time for disruptive technologies to snowball!

Inflection points we left behind (1) Voice & Data Capacity in Mbps Source: Mutooni & Tennenhouse, MIT, Jan 1998

Inflection points we left behind (2) 1 10 100 1,000 10,000 1986 1988 1990 1992 1994 1996 1998 2000 2002 Aggregate Bandwidth (MB/s) Ethernet Internet Backbone 1 Gbit/s OC 192 T3 I/O Buses 10 Gbit/s

Global Internet Hosts (000s) 1989-2006 360 Million users Source: Vint Cerf, MCI Worldcom, Jan 2000 Inflection points ahead of us (1)

Incomplete transformation; the inflection point is quickly approaching … Cisco Nortel Juniper 3Com Network & Mgmt services Embedded OS System ASICs ? ‘ 00 Vertical Network Industry Horizontal Network Industry Inflection points ahead of us (2)

2000’s Thin Client Thick Server 2001’s Thin Client Smart Network Server Farms Data Farms Inflection points ahead of us (3) 2001’s Thin Client Thick Server 1996-2000 Thick Client Thick Server 1970-1985 Dumb Terminal Mainframe 1986-1995 Thick Client Thin Server

Kiss “The Network” Goodbye!!! How about a “white-sheet” network with nodes exposing programmable control to 3 rd party code? How about passing on to entrepreneurs and service providers the freedom to formulate the high pay-off services for which they have found a market For this, we will need new levels of abstraction

Fred Smith invented the FedEx business in 1973, after recognizing the significance of predictable movement of packages. Smith has always used the same roads, airports, cars, and types of planes as the U.S. Postal Service … but he has programmed his own resources to work in a novel way Much like …

terminals in airports, railways, roads fleet, personnel parcels Analogy APIs and resources in telephones, routers, switches Ubiquitous, shrink-wrapped software PDUs (meeting the various SLAs)

Service sampler Entrepreneurs will run code inside the network for: Ad-hoc routing or policy definition for VPNs New protocol versions or features deployment Multicast protocols and support tools instantiation Stateful packet capturing; “intrusion detectors with legs” Intentional directory services Content-sensitive load balancing and caching Diagnostic agents (e.g., for loops, SLAs infringements, etc.) Customizable IP Accounting (e.g., to support “pay-for-what-u-use”)

Why inside the network and not e2e? 1) Brake! 2) ABS detect lockups and will pump brakes ABS brakes are a Turing machine exploiting locality

Why inside the network and not e2e? (2) Fan-in/fan-out Centralization/De-centralization Visibility Affinity with data that only lives in the network Localization

Location, location, location Service-enablement will prove most effective where “impedance mismatches” occur in the network Wireline vs. wire-less Secure vs. non-secure Customer-premises vs. Content-provider-land SLA (x) vs. SLA (y) Low TCO vs. High TCO Resource-constrained vs. unwashed unlimited computing A service-enabled box can wear multiple hats

Where have you gone, network layer? 1 2 3 old world routers Physical Link Network Transport Session Presentation Application 1 2 3 4 5 6 7 Physical Link Network Transport Session Presentation Application 1 2 3 4 5 6 7 Physical Link Network Transport Session Presentation Application 1 2 3 4 5 6 7 Physical Link Network Transport Session Presentation Application 1 2 3 4 5 6 7 new world edge router (few) 1 2 2.5 routing switch (many) 1 2 2.5 routing switch (many) 1 2 3.5 4 5 6 7 proxies

Emancipation of a router It all started from old-world, vertically-integrated code ASICs/Processors Proprietary Apps Proprietary NOS

1 st Degree of emancipation Introverted APIs emerge Modular code is native, local, and trusted . WAT port required ASICs/Processors Forwarding Engine S y s t e m S e r v i c e s F r a m e w o r k Routing Protocol 1 N Routing Table Manager Forwarding Engine Interface System Manager Management Interface Agents 1 N M M M N O 1 O N C C FC 1 N FM

2nd Degree of emancipation Extroverted APIs expose object capabilities to ISV code ISV code is a local, native, and trusted .DLL; loaded WAT Introverted APIs APIs ASICs/Processors Forwarding Engine Extroverted APIs ISV’s Software S y s t e m S e r v i c e s F r a m e w o r k Routing Protocol 1 N Routing Table Manager Forwarding Engine Interface System Manager Management Interface Agents 1 N M M M N O 1 O N C C FC 1 N FM

3th Degree of emancipation Extroverted APIs extend a commodity Java runtime ISV code is local/non-local, non native, non trusted, and is loaded on demand Extroverted APIs Introverted APIs APIs ASICs/Processors Forwarding Engine JVM JAPIs ISV’s Software S y s t e m S e r v i c e s F r a m e w o r k Routing Protocol 1 N Routing Table Manager Forwarding Engine Interface System Manager Management Interface Agents 1 N M M M N O 1 O N C C FC 1 N FM

4th Degree of emancipation ISV code is local/non-local, non native, non trusted, loaded on demand, and can teleport itself Extroverted APIs Introverted APIs APIs ASICs/Processors Forwarding Engine JVM JAPIs ISV’s Software S y s t e m S e r v i c e s F r a m e w o r k Routing Protocol 1 N Routing Table Manager Forwarding Engine Interface System Manager Management Interface Agents 1 N M M M N O 1 O N C C FC 1 N FM ISV’s Software ISV’s Software

Top 5 Challenges to Service-enablement What do the APIs look like Scalability Service guarantees Security “ I’ve got a hammer and everything looks like a nail”

Who’s looking into this? Programmable Nodes community Active Networks community Akamai-like infrastructures Jain Intel’s Phoenix platform Lucent’s softswitch Nortel’ s openet.lab platform

Programmable Nodes Current Nodes Vertically integrated Primitive scripting Programmable Nodes Manufacturers publish extroverted APIs I can port code and extend such network nodes Alternately, I can operate the APIs remotely APIs’ target audience: Manufacturers, ISPs References IEEE P1520 Industry alliances

Active Networks Current Networks Data in the packet, program in the network nodes Effective, but inflexible Active Networks The program travels with the packet, end-to-end Flexible, but ... how about security, interoperability? Most ambitious thus far Target audience: ISPs, end-users References DARPA Active Networks

Partial overlap Active Networks Programmable Nodes Users extend routers Manufacturers extend routers ISP extend routers

Nortel’s Openet.lab It’s an incubator for service-enabled network nodes and sample services It provides: JVM-emancipated prototypes of Nortel routers Java APIs to MIBs Java APIs to Forwarding Planes, packet capturing A runtime environment for downloaded code It’s popular among researchers (DARPA, CSIRO, …) Free downloads from http://www.openetlab.com

Service-enabled Network Flashcards (1) Client Client Client Client Client Client Client Object Ref. Caching Obj. Load-balancing Connection Pooling Preferential QoS Access Policy Push for DB Sync. Stockwatch Client Presentation Objects Stockwatch Server Biz logic Objects Intranet Stock-history DB Data Objects Real-time Database Internet Object Directory

Service Provider: - Directory - Bank - Etailer - ASP Service-enabled Network Flashcards (2) Content Provider Internet P S T N Internet

Service-enabled Network Flashcards (3) Billing Auditing Planning Surveillance Census Accounting Applications higher goodput higher goodput Accounting Server Network Nodes Activated edge nodes are taught to prep accounting data and peer ad-hoc accounting protocols

Conclusion Fast-paced innovations paint an optimal scenario for service-enablement in the network Service-enablement is an effective way to overcome network “impedance mismatches” It brings friend-services and friend-content closer; it pushes foes further away Kiss The Network good bye. It’s My Network now!

Closing remark Back then, thrust wasn’t a problem; control was Likewise, network bandwidth growth is just fine; control does demand our collective efforts

Active network

More Related Content

What's hot (19)

Similar to Active network (20)

More from Michel Burger (6)

Recently uploaded (20)

Active network

Editor's Notes