Over the next few years enterprises that are looking at the next generation of network services will be able to free themselves from the tyranny of vendor lock-in, the days of requiring costly hardware upgrades to realize new network capabilities will be over.
Major technology platforms tend to last about 25 to 30 years. Change happens when the shortcomings of the dominate platform become too much to bear — as with the Mainframe, Hierarchical Databases, and more recently the PC. With the change a new platform emerges to take the old platform's place and bring the future forward.
Current networking topologies have not kept pace with the transformation to virtual services, enhanced WiFi, Bring Your own Device, (BYOD), etc., and instead networking is bogged down in a 20-year-old operational model designed for manual provisioning and protocols built to only support device-by-device networks. Because of these issues networks are overly complicated, fragile systems constructed from hundreds of individual devices tied together by complex and vendor-specific protocols with no programmatic solution for network-wide control. This is why the data networking architecture of the past 20 years; characterized by smart but propriety architectures that are closed and expensive has finally run its course.
What the network will start to look like is something you are already using; cellular service. How network architectures will start to look and act like the cellular network in the next few years are:
End User Network:
Wireless versus todays copper LAN.
Users will connect over WiFi at greater speeds than the current copper LAN based speeds: WiFi 1.3Gbps versus Copper 1Gbps.
All user traffic across the WiFi network can now meet NSA Top Secret communications standards allowing agencies to transmit without the risk of WiFi intrusion: WiFi secure, copper unsecure.
At airports and other congested WiFi locations the elimination of WiFi channel conflicts with the growth in the number of WiFi channels from 11 to 160: WiFi available to all providers securely without channel interference.
Data Center Network:
10/40/100 Gigabit Ethernet, (10GbE, 40GbE, 100GbE) versus 1Gigabit Ethernet, (1GbE).
Servers will be moving from physical platforms with 1GbE connectivity to virtual servers with 10GbE connectivity.
Switches moving from 1GbE to the servers and multiple 1GbE and some 10GbE connectivity to 10GbE at the server and 40GbE and 100GbE between servers and network application appliances, (firewalls, IPS, WiFi Controllers, etc.).
Network security and application service devices moving from 1GbE to 10GbE and 40GbE. To support the demands of virtualization network appliances like firewalls will need to support 40GbE speeds and throughput to keep up with the virtual server and end-user requests.
The magic of cellular service: call from anywhere to anywhere.
For cellular what makes it all work is the database: Signaling System Number 7, (SS7) developed by Bell Labs and then Telecordia in the 1980's. SS7 replaced traditional switching topologies that would not allow for anywhere - any time or enhanced services like SMS , (text messaging). SS7 solved the anywhere-to-anywhere problem while lowering OpEx and CapEx per call.
For data networks the SS7 revolution is built around Software-Defined Networking, (SDN). Like the cellular carriers; agencies using SDN can decouple the physical proprietary equipment from the networking software adding functionality to the network on any vendor's equipment while lowering capital costs and operational expenses by commoditized network equipment using SDN services.
Is this really happening? Juniper purchased in late 2012 Contrail Systems for $176 million. Contrail is a SDN developer. A company with 50 employees started less than a year ago, and no product just an idea on how to develop a SDN software suite. Yes SDN is the future of data networks just like SS7 is for cellular.
Enhanced WAN Services:
For most Internet access, the preferred method is cable and DSL. Over the last few years cable speeds have increased to 100Mbps for $170 a month, and ADSL, U-Verse, FiOS to 30Mbps and greater. Copper T1 services first deployed in 1955 are slower and more expensive than a Cable modem + DSL access and at 1/3 the price of a T1. New WAN services include:
DOCS-3 for cable services are offering up to 100Mbps for access.
ADSL offers speeds up to 30Mbps.
U-Verse up to 100Mbps.
FiOS up to 300Mbps.
Ethernet over copper 45Mbps.
Ethernet over MPLS up to 10Gbps.
All these services at prices less than traditional T1 and T3 services allowing for redundant unique pathing, (Cable+ADL to each location) to ensure availability if one vendor is down.
The time is now for evaluating new network services
Current enterprise architectures are a barrier to achieving the promise of virtual computing and enhanced services to the end-users. Most enterprises are bound to physical network hardware and topology. This binding results in limited scalability, increased complexity and inflated costs. A stopgap approach has been to upgrade the existing equipment to new equipment, but this approach does not fix the underlying problem and, in most cases, requires expensive hardware upgrades and binds network operations even more tightly to one vendor without enhanced performance.
The goal of the 2013 network architectures should be that a service can be introduced without displacing the old platform. Changes include:
IEEE 802.11ae for wireless access and services to allow for 1.3Gbps to the end user versus land locked copper services.
Supports three bandwidths: 2.4GHz, 3.7GHz, (new), and 5GHz.
NSA Suite-B Security for secure WiFi transmission.
WAN-II services like Cable, ADSL, FiOS, etc.
Continue to support legacy 1GbE.
Use 1GbE for Out-of-Band service, Backup and Recovery, Management, etc.
Migrate production to multiple 10GbE with LACP binding multiple NIC's to multiple upstream switches.
Multi-Chassis Link Aggregation, (MC-LAG), offering highly available server to switch topology.
SDN services across all network elements and applications. Having an application like Oracle advising the firewall of an intrusion and asking the firewall to block the traffic. VMware advising the network of a migration and requesting a best path for the minimum conversion across the data center or WAN.
Virtual switching that supports SDN.
Switch to switch at 40GbE and 100GbE. With SND providing best most available path across the architecture.
Migration to flatter network topologies. Moving from traditional three tier hieratical networks to two tier Spine-Leaf networks for faster performance and fewer hops between components.
SDN for applications and network elements. Making the operations teams and the network itself aware of end-user and resource requirements balancing the network architecture to meet anytime demands.
Explore the deployment of cable+ADSL at Cat-4, Cat-3 sites for lower costs and higher performance to the data centers and regional services.
Over the next few years enterprises that are looking at the next generation of network services will be able to free themselves from the tyranny of vendor lock-in, the days of requiring costly hardware upgrades to realize new network capabilities will be over. The next generation of network services will create an intelligent network edge managed by a control cluster that transforms the existing physical network into an IP backplane and enables the programmatic creation of thousands of agile networks to connect workloads anywhere.