The Workload Placement Strategy comes first
Broadly, there are two starting points for the cloud journey:
- Next generation outsourcing- using IaaS and PaaS to lower the cost of IT
- Digital transformation- improving the business by writing and iterating software more quickly
The right reasons to go to the cloud generally derive from the business – improving time to market; enabling new product and capability; entering new markets; a preference for opex over capex; reskilling a finite team to higher-value functions.
An organization needs to settle upon an approach, sometimes expressed as “cloud first” or “last”, “lean toward” or “against.” Yet whatever the high-level approach, there are exceptions.
On the one hand, an organization may move to cloud because there’s a “wrecking ball” event on the horizon: an internal data center needs to be repurposed, or requires a significant capital outlay, a colocation lease, a lease for hardware or a software license is coming to term.
On the other hand, cloud is not for every workload. At a recent client, a mission-critical application was characterized as “a thick-client 2-tier architecture that has embedded affinities to the HP-UX operating system…current state platform constrains migration options.” Further, its “Relational Database Management instances are deployed in physical environments – some are single points of failure.”
Beyond technology and architecture, process can impact the timeline. At this client, performance and availability ran at a very high level- yet the operation was not particularly well documented. Success based on institutional knowledge will not immediately translate to a new environment. Resiliency must be architected from the beginning, and the processes documented.
So while the high-level strategy was broadly “cloud smart,” it would clearly take the organization some time to move substantial elements of the application suite, and the financial business case was complicated by the need to run parallel infrastructures while beginning to acquire cloud expertise.
Colocation is the right approach for most organizations
A combination of demand and supply-side considerations drive the changes in delivery models. First, cloud has created uncertainty in data center demand: workload profiles will change rapidly and substantially over a 3-5 year period. And secondly, led by the early adopters, the colocation value proposition has expanded from traditional “ping, power and pipe” to colo at the center of their connectivity strategy.
Several providers invested early; more are following each day; and for most enterprises, the increasing demands for connectivity are best met at the scale of these operators for whom cross-connection is increasingly table stakes.
As expertise grows in managing the colocation providers, selection criteria change.
- Simplicity vs. transparency: while all-in price per KW is convenient to manage, for a largely idle DR site, a metered approach may be better than a flat rate; the goal becomes transparency
- Capacity negotiations: previously focused on adding contiguous space, clients are now frequently focused on the flexibility to downsize. They’re right to do so: many clients report significant overprovisioning.
These changes drive an evolution in process and selection criteria:
- Selection team expands to involve enterprise architecture or business applications, network specialists, security teams and continuity specialists
- Providers must demonstrate ongoing commitment to updating investment
- Manageability considerations are important: ideally, migration is timed with refresh cycles, substantially reducing the risk and cost of swing gear.
- Strategic procurement preferences for single vs. multiple providers
- Contract terms seeking equilibrium: the end user wants flexibility and the provider wants commitment. A balance must be found
Increasingly, the colocation provider decision is based on connectivity as much as space and power
GTSG has referred to space and power as “colocation value proposition 1.0”, with the enhanced value of connectivity, cloud interconnect and onramp adding layers to the value they provide. As noted above, for many organizations the scale on which these operators provide will overwhelm any economies that an individual enterprise can architect or negotiate.
As the connectivity value proposition increases in importance, evaluation criteria can be weighted toward interconnect and on-ramping. For example,
- Connectivity: carrier neutrality; options; rapid provisioning
- Network Equipment: IPv6; GigE support; IP address block assignments
- Network Services: SLAs (four, five, six nines); web load balancing; DDoS mitigation; disaster recovery
- Cloud enablement: migration services; private cloud and connectivity, cloud on-ramping services, self-service multicloud connections
Several colocation providers have invested aggressively in connectivity, ahead of broad end-customer adoption. The investments of these industry leaders are well documented. However, the marketplace changes continuously both from new provider investment and by provider consolidation. An individual use case may not warrant the price premium associated with one of the interconnection leaders. We recommend, and assist with, an RFP at the appropriate time. Critically, we also get underneath the workload placement strategy in a way that brings clarity to the connectivity requirement.
GTSG has been assisting clients with infrastructure delivery strategy – and the migration efforts which make the strategy real – for over 20 years. If you’d like to discuss your evolving Infrastructure Delivery Strategy, and the options available to you through a changing colocation marketplace, please write us at Partners@GTSG.com. Thank you.
 https://www.globenewswire.com/news-release/2020/05/25/2037962/0/en/Data-Center-Colocation-Market-to-Grow-with-14-8-CAGR-in-Coming-Years-P-S-Intelligence.html, retrieved 04.19.21