Data Center Migration
Hey Friends, Nerds, and Geeks, we are back today with our Audio podcast show # 25 and its a Guest Expert show! My good friend Malcolm Booden from the UK Joined me in Jan for a discussion around Designing for Data Center Migration and Application Mobility. We talk about a lot of different concepts, perspectives, and items to keep in mind in when you are going through an event like this. Today is the first part of the discussion. So sit back, turn up the volume and enjoy the show!
Data Center Migrations We are going to talk about data center design and more specifically design when migration of Data Center services either to a private or public cloud are considerations. What should you think about when performing a Data Center migration of any magnitude, and what is the thought process that has to come with such an initiative?
What is really involved in planning a Data Center migration. No matter what specific type of Data Center migration you’re performing, the main things to understand before you start the design are:
The Five Main items to Consider
Business drivers and constraints
Applications in scope
WAN topology
Traffic Flows (users and apps)
What is in place now?
Data Center Migrations vary in size and complexity Data Center migrations can vary in size and complexity and be very different depending on what the objective is, but we can touch on a few different Data Center migration types and what may drive each of them:
First of all there’s different scope and scale of a Data Center migration depending on what is happening. Large scale migration of tens to hundreds or thousands of apps, or is an application being moved to a new location i.e. SAAS cloud, hosted etc. For now, we are going to concentrate on “full” Data Center migrations where multiple apps need to be migrated.
New applications or services being deployed in a new Data Center for specific purpose i.e.
Cloud app, private cloud app
Micro-Clouds
Micro-Services
More of an application migration
Data Center migration could be replacing critical infrastructure such as the Data Center core switching architecture, like for like or to a different platform i.e. to new vendor or introducing an overlay to deliver Software Defined Data Center (SDDC) where applications are remaining unchanged.
Moving applications from one Data Center to another for one of the following reasons:
Relocation of office
Moving services offsite / colocation (space, power)
Managed service award to outsourcing company
Key Requirements and Drivers
Stand up applications or services in a new location
Acquisition and mergers, managed service outsources – Enterprise
End of existing managed service and moving to a new provider – Enterprise
New services being offered by a provider – MSP / ISP
Constraints and Challenges
Cost
Time to deploy / market
Skill sets available
In house staff
Outsourced
VAR availability
Other Items to consider
IP Addresses (PI and PA)
DNS
Switching
Firewalls
Remote Access
2FA
Load Balancers / Local DNS Global (Anycast, DNS TTLs etc)
WAN connectivity (Overlay vs Native L3VPN etc)
WAN Performance / Optimization
Voice / Video and performance especially in global networks
New or Lift and shift applications – or both
Impact this migration has on addressing (Re-IP applications usually not feasible)
Normally IP addresses within code
Re-IP exercise may be completed to tidy up noncontiguous DC ranges and migration time may be best time to consider
Legacy applications which are sensitive to TCP drops require
Consider firewall clustering
Knock on effects of L2 stretched between sites
Ecommerce requirements, persistence etc for 24/7 services
Availability / downtime available
Layer 2 (or 3) DCI – whole show in itself!…….why we should or shouldn’t do L2 DCI, but how it can be achieved!
Software Defined Data Center (SDDC) We are now beginning to enter the SDDC territory so the question is raised how and when do we migrate to software defined architecture and is it feasible for everyone?
Different Technology Options Example: LISP + OTV – where OTV was the main requirement How to route LISP using tunnels or redistribute into existing WAN routing protocol LISP Multihop with tunnels over WAN, Firewall in path between XTR and EID / FHRP router. Service Provider core had routing limit on number of routes in VRF where injecting large number of LISP host routes would have broken the network (potentially). This is often the challenge – seeing what may happen if something is designed a certain way in advance.
Example: DNS load balancing such as F5 if the budget was available Budget is most commonly the factor which plays into “being creative” – that said with something like a Data Center migration there is normally a decent budget for the purposes of that organization but requirements always come out of the woodwork within the project and it is when unexpected factors which cost money appear that is when you sometimes need to be creative.
It is about finding a balance between something that will deliver the solution required, but is also supportable.
The Term of the Show
Data Center Interconnect (DCI)
Call to Action:
Share this content with someone you believe would benefit from it as much as you have!
Hosted By: Michael “Zig” Zsiga
Guest Expert: Malcolm Booden
Round Trip Technology – Blogs, Vlogs, and Audio Content
Twitter!!
Thank you for the podcast reviews! Thank you for the iTunes reviews as they sincerely motivate me to continue to give back to this community that I love. If you enjoy the podcast and the information I share, please feel free to write your own personalized review on iTunes!
Ask questions and give feedback
You can leave a comment in the show notes
You can leave a voicemail at (617) 913-4103
You can email us at Feedback@zigbits.tech
Engage with Zigbits further:
Subscribe to the podcast on an iPhone or on an Android
Follow Zigbits on Twitter!
Follow Zigbits on LinkedIn!
Follow Zigbits on Facebook!
Engage with me further:
Follow me on Twitter!
Follow me on LinkedIn!
Transparency: This post may contain affiliate links to products or services were I may receive a level of compensation from your actions by following those links. This is seamless to you and does not add any additional cost to the products or services in question. In addition, I do not let any affiliate relationship cloud my judgement or my recommendation of a product or service. My recommendations will always be above reproach. This is my commitment to each and everyone!
The post ZNDP 025 – Designing for Data Center Migration and Application Mobility with Malcolm Booden – Part 1 appeared first on Zigbits - Where Zigabytes are faster than Gigabytes.