Carrying on from my thoughts in Part 1: It’s time to start deploying purely standards-based infrastructure outside the data center; data center protocols are just starting to be created for a converged and virtualized world. With the amount of tested and deployed standards protocols, there’s no excuse for networks to be locked in to a certain vendor with proprietary protocols when standards-based network solutions provide access to compelling volume economics, flexibility to adapt a much wider array of solutions, and relief from hiring specialized talent to run a science project. Although many organizations understand that standards-based networking provides them with the flexibility to choose from the best available solutions at a lower cost of ownership, they still feel trapped. Listed below are three top shackles and the keys to open them up:
- “We are using advanced features like EIRGP and PVST+; we aren’t sure there is anything similar to accomplish the same task.” Whether using Cisco Discovery Protocol (CDP), Juniper JFLOW, Brocade’s VRRP-E, or other proprietary features, the standards bodies (IEEE, IETF, ISO, etc.) have done a lot of work in the past 10 years and provided similar functionality with LLDP, SFLOW, OSPF, VRRP, as well as others. Some of the standards are more sophisticated than their proprietary brethren. Most of them have been around for awhile and are being utilized in the most demanding environments like CERN (birthplace to World Wide Web) and Lawrence Livermore National Laboratory. Anyone would be hard pressed to show that a network could not be created on standards that wasn’t equally secure, reliable, low latency, high throughput, intelligent, and efficient compared with one implemented with proprietary protocols.
- “We have invested a large amount of time and money having our personnel receive Cisco certifications (CCNA, CCIE, CCNP).” Please don’t confuse certifications with proprietary implementations. Cisco has created one of the best network training programs in the world; the courses spend 90%+ teaching about good networking architectures, best practices, and protocol theories rather than specific Cisco proprietary features. For example, students spend a lot of time learning about spanning tree before they learn how to implement per vlan spanning tree plus (PVST+). This fundamental knowledge easily provides a baseline, not only distinct features from Cisco but networking protocols in general. Anyone who has taken the courses and passed the related exams are 99+% on their way to supporting standards-based infrastructure. The worlds are so close that vendors provide “accelerated tracks” for Cisco-certified personnel. After hundreds of hours in getting a top level certification at Cisco, it only takes four hours afterwards to get HP certification.
- “We don’t have a lot of time or money to migrate from a vendor’s xxx protocol to standards-based protocol.” There are one-time costs that cannot be avoided, but they can be minimized if it’s done when network refresh is required. These challenges are in line with similar infrastructure projects like new storage, middleware, and application platforms. The migration from EIGRP to OSPF has more benefits than drawbacks when it comes to the ownership of the customer’s infrastructure. This is a mature, well understood process now. The actual migration from protocols like PVST+ to MSTP or EIGRP to OSPF, although being very well documented and fairly easy to perform, may require your staff to review their certification literature for a quick refresh of the details or hit up a vendor who has technical documentation around the process.
Overall, enterprises who have moved their networks over indicated that the upfront hurdles associated with the switchover were less than expected, and that the benefits of avoiding lock-in, leveraging skills from a more widely supported population, and architectural flexibility were more liberating than expected.