|
Juniper adds new dimension to router front
After a successful performance in the core routing space,
Juniper is now eyeing the edge of service provider networks. It is also looking
at the Model for Integrated Network Transformation (MINT), which allows disparate
networks to converge on a single network. Rahul Neel Mani has the details on
Junipers plans
 |
With Junipers 10 G-enabled multi-service
edge platform users can move on to the newer emerging Layer 2/3 services
in the future, says Giridhar Java |
JUNIPER Networks, a company threatening Ciscos monopoly and a powerful
competitor to others in the core routing space, is looking beyond core now.
Until recently, Juniper was perceived as just a core routing company where,
in the last one year, it has grown by nearly 30 percent, with market share increasing
from 16 to 21 percent. In the last quarter alone the company has grown from
18 percent to 22 percent, registering a smart 25 percent growth, whereas other
legacy network equipment vendors have actually seen sales falling.
Juniper is now gearing up to capture the edge of service provider networks.
The company unveiled an edge router called M320, designed to meld Layer 2 services
(such as Frame Relay and ATM) with an IP/MPLS backbone. Along with the router,
it also rolled out a software toolkit to allow service providers to write applications
for Frame Relay and ATM service emulation. The toolkit, called J-FASE, also
fosters interoperability with Lucents Frame Relay and ATM switches (Juniper
and Lucent have an alliance centered around migrating Lucents installed
base of Frame Relay and ATM customers to MPLS cores). The M320 Multiservice
Edge Platform is designed to consolidate points of presence (PoPs) by replacing
legacy routers. This new router now pushes Junipers two-year-old M40e
edge router further down into smaller PoPs.
M320: Junipers bet for the edge
The M320 incorporates a new routing engine, the 1.6 GHz/2 G DRAM RE1600, which
is designed to support thousands of VPNs and customer interfaces. The router
features a forwarding rate of 385 M packet/sec and a switching capacity of 320
G bit/sec. Interface densities are 16xOC-192, 16x10 G Ethernet, 64xOC-48, 128xOC-3/12,
160x1 G bit/sec Ethernet and thousands of DS-1s, DS-0s and logical interfaces.
The M320 also supports eight queues per logical interface for Quality of Service
(QoS). The M320 supports Metro Ethernet services like VPLS, per-VLAN QoS and
stacked VLANs, as well as translational cross-connect, a Juniper-developed feature
for ATM/frame relay-to-Ethernet interworking.
The J-FASE includes capabilities such as DiffServ-aware MPLS traffic engineering
and connection admission control for guaranteed bandwidth across IP/MPLS networks.
It also includes operations, administration and maintenance tools to enable
providers to assure service-level agreements.
Giridhar Java, country manager, India & SAARC, Juniper, says that the company
already had a suite of M-Series products. After listening to customer feedback,
it felt that there were islands of networks for multiple layers of services
that a service provider (SP) markets. Each time a SP thinks of a new service/revenue
stream, it has to build a new network infrastructure and add another layer over
the old one. It results in very high capital expenditure (capex), and
because of multiple boxes and platforms, the manageability of the infrastructure
becomes a big issue. This results in increased operational expenditure (opex),
says Java. Juniper started off with the M-Series to allow for various avenues
on a single platform which was based on IP/MPLS, and then moved to the MINT
(Model for Integrated Network Transformation) philosophy, which further allows
for these revenues to turn into profits.
M320 Multiservice Edge router is trying to connect the POPs by replacing legacy
routers. It pushes its own legacy M40e series edge routers into the smaller
PoPs. Java explains how it happens. If a SP has invested in STM16 ports
and a Gigabit Ethernet box and has positioned one box for every network, then
there are several such networks within one SP network, resulting in scaling
hassles. Even if it scales, performance is impacted, he says. When the
SP moves up to the next step, should he essentially keep investing in newer
networks thus creating multiple capex? What is the solution? The M320
is the convergence point for all technologies. The SP moves into a single infrastructure
environment that allows you on to the big box through an IP-based MPLS function.
There can be tremendous reduction in investments made in multiplying networks/numbers
of boxes, Java explains. This allows disparate networks to converge on
a single network. Having said that, what is also important is security, performance
and scalability. There should be no degradation in performance when you have
to scale up the network. In case of a M320, it has been further enhanced with
a new 1600 Routing Engine (RE) based on the IP2 ASCIS, including the M40E. When
you are pushing M320 on the edge for supporting multiple services, then you
can put M40E further down the network where it can be best utilised.
M320 provides a separation of forwarding plane from the control plane. It can
provide eight times the capacity in forwarding plane that an M40e could give,
and 2.5 times the control plane capacity of M40e. Juniper has the industrys
most reliable, scalable, 10 G-enabled multi-service edge platform. It is a strategic
investment so that you can move on to the newer emerging Layer 2/3 services
in future, says Java. There are more than six Layer 2 services that service
providers can carry today, but with time there will be many more services. Services
like VPLS, IPv4/IPv6 etc, will become quite prevalent. Thats where
we are talking about, not just pushing boxes but to protect investment with
more services. All this while we have been stressing on service reliability
and protection, now we are talking of service protection, says Java. Junipers
M320 allows for a greater level of service protection because of its support
for current and emerging Layer2/3 IP/MPLS-based VPN services.
M7i & M10i vs Cisco
Last November, Juniper also unveiled two edge routers designed to unseat older
routers from arch-rival Cisco that are already installed in service provider
networks.
Called the M7i and M10i, these routing platforms are targeted at small and
midsize PoPs, campus networks and managed service environments that might be
straining the performance and feature limits of Ciscos 7200 and 7500 series
of routing platforms. Juniper claims that the new routers let service providers
turn up any feature or service without compromising performance; the company
contends that Cisco routers cant offer the same. At four slots and two
rack units, the M7i chassis is Junipers most compact M-series router.
It features integrated ASIC-based J-Protect security, which includes network
address translation, Stateful Firewall, IP Security and J-Flow accounting.
The M7i, supporting 16 million packets/sec performance, can serve as a provider
edge router in small PoPs and customer premises equipment systems for managed
services and campus border router applications.
Juniper claims that Ciscos upgrade to the 7500the 7600 Optical Services
Routeralso suffers from a service/performance compromise, and is missing
some key MPLS Provider Edge functions. The company is hopeful that service providers
will either yank out their 7200s and 7500s or push them down the network for
the new Juniper routers even though the rest of their infrastructure is Cisco-based.
Java admits that 6-8 months ago Juniper was present in just 9-10 of the top
25 service provider networks of the world. But as of date, Juniper is in 24
of them. Those service providers were then having nothing but legacy equipment.
So what happened to that? The whole point is that people dont replace
infrastructure but push it down the line. People push it towards the edge and
subsequently towards access, says Java. When customers saw that
they could take more value-added services to the market, they positioned the
infrastructure at the top with Juniper making it a core routing company interoperating
with other network equipment. But when the company launched the Junos and the
M Series, the new architecture-based routers, we were destined to become the
edge router company, states Java.
Although its not a common occurrence for service providers to replace
Cisco routers with Juniper platforms, the market is very unpredictable. And
surely, dislodging Cisco will prove very tough for Juniper. Service providers,
looking for a raw bandwidth boost might be inclined to evaluate the Juniper
routers as replacements for the Cisco platforms. But the edge is where feature-rich
services are defined and tuned. Its also where service providers make
their money.
Infranet will be the decisive battle Resulting out of MINT model, Juniper
has now started talking about the Infranet
initiative. The Infranet initiative seeks to develop two interfacesa
user-to-network interface between customers and service providers and a so-called
inter-carrier interface between service providers that adheres to a set of interconnection
standards that establish a lowest common denominator required for
implementation.
Infranet is not just about a standard or proprietary technology talked
about by Juniper. Its a call to action by industry vendors and service
providers to come together and provide a similar experience, says Java.
It endeavours to build user-to-network and network-to-network
interfaces just like inter-carrier exchanges between service providers. It
should enable any service provider to seamlessly carry voice/video/data with
the service level agreement (SLA) that has been signed between any vendor with
any user. Its just like the roaming service for mobile service users.
This inter-carrier agreement would become the Infranet, explains Java.
But this is not a breakthrough initiative by Juniper alone. It has been talked
about in past by other vendors too, but nothing much happened about it. Juniper
argues that the technology is here now; whats missing is collaboration
in the industry on how best to implement it to achieve this application-aware,
service-granular Infranet. Current peering relationships between service providers
and service-level contracts between customers and service providers fall short
of the Infranet initiative goal because service assurances cannot be guaranteed
across peering arrangements and certain types of traffic cannot be granted a
greater level of treatment. What Juniper is proposing is selectively
open connections between carriers, defined by common standards such as
MPLS, Resource Reservation Protocol and the Layer 2 Tunneling Protocol, that
support and reward the delivery of advanced services, such as content distribution
and virtual private networks globally, vs a carriers own physical network,
says Java.
Juniper hasnt provided a timeframe for completion of the Infranet initiative.
Juniper is already working with vendors like Lucent and Microsoft and many others
to help it reach its objective. But some analysts call it just a vendor-motivated
initiative that ends up selling proprietary network framework. But Juniper refutes
the charges. The company says that had it been a Juniper-centric development,
there was no reason why companies like Lucent, Microsoft and Global Crossing
would take part? Its slow to catch on. It will take time for people
to take it but things are surely moving. Infranet initiative will be completed
in three phases. The first phase is to build the platform infrastructure to
carry the traffic. The second phase will be the controls, policies and management.
The third phase will be to build the user-to-network and network-to-network
interfaces. The first phase is currently on, says Java.
rahul@expresscomputeronline.com
|