As a multi-tenant data center provider, CoreSite serves customers ranging from global enterprises to startups. Some come to CoreSite knowing exactly what they want. Their IT teams are experienced with everything from cross connects to hybrid cloud architecture and Ethernet Virtual Circuits. Many others, who are new to colocation or have been running operations through an on-premises data center, really benefit from talking through options with a CoreSite Solutions Architect (like me!).
But what does that conversation typically entail?
In this post, I’ll unpack how we guide those decisions, starting with a few key points relevant to the pros and cons of on-premises data centers, multi-tenant colocation data centers and the public cloud.
On-Premises vs. Colocation vs. Cloud Data Centers
Enterprises are in the midst of a “good problem.” There are many options for distributing traditional and AI workloads, storing data and fine-tuning interconnected workflows. Solving the good problem starts with understanding which data center environment is best for each element in an infrastructure.
On-Premises
Not every workload belongs in public clouds, and the shift from “cloud-first” to careful repatriation is well behind us. On-premises data centers are where digital transformation began and still have a role to play.
Advantages of owning and managing your infrastructure include:
- Customization – You configure the servers and networking based on specific requirements.
- Lower Total Cost of Operations – Initial investment into equipment could be significant but costs could be lower over time.
- Data Control – Data is under your management and protection.
Drawbacks of on-premises data centers include:
- Management and Maintenance – Dedicating resources to keep equipment, systems and software is a line-item expense; skilled personnel could be required to “keep the lights on” as opposed to developing.
- Scalability – Unpredictable lead times for hardware, capital outlay for equipment and even building out the footprint are scalability roadblocks.
- Risk of Downtime – Many enterprise data centers are about 10 years old, not “purpose-built” for modern workloads and designed for resilience.
Public Cloud
Multi-cloud deployments are the norm these days, with enterprises rolling public cloud services for executing core business processes, storage and AI model training into the mix … for good reason. As a network engineer and solutions architect, I’ve witnessed how “the cloud” served as the resource of choice for SaaS applications, data archiving and, most recently, generative AI development and implementation.
The advantages of public cloud data centers include:
- Low Up-Front Expense – With no hardware investment or facility to operate, initial costs are low.
- Flexibility and Scalability – Myriad services offered through the variety of cloud providers can be easily initiated and terminated; 100% virtualized, increasing the amount of processing power is easy and welcomed.
- Remote Accessibility – Public cloud gives you the “anytime, anywhere” access you read about all the time.
Drawbacks of public cloud data center use include:
- Cost Control Challenges – Although many services are offered as “pay by the drink,” it’s difficult to gauge usage, leading to surprise cost increases. Data egress fees can be high.
- Vendor Lock-In – Public cloud business models can be designed to make it difficult to exit contracts.
- Latency Issues – Modern workloads can require real-time or near real-time data transfer latency that public cloud can’t deliver.
Colocation Data Centers
Before public cloud matured, growing businesses needed to extend beyond their on‑premises data centers. What did they do? They turned to carrier hotels where network providers were peering with one another to “co-locate” the equipment rapidly becoming the lifeblood of commerce and communication.
Colocation is an extension of on-premises – without scalability constraints, significantly lower risk of downtime or network limitations – and a bridge to the cloud.
The advantages of colocation data centers include:
- Business Continuity – Redundant power and cooling systems, diverse network connectivity and a staff on-site 24x7 to keep systems operating uninterrupted.
- High-Density Support – Power, cooling (including liquid cooling) and low-latency data transport critical to modern workloads and, of course, AI.
- Digital Ecosystem Interconnection – Rapid, agile access to the network of organizations, platforms and services that exchange data essential for a business to operate, including direct on-ramps to public clouds.
Before I spell out some disadvantages of colocation, I’ll point out that not all colocation providers can deliver the advantages listed above. Which is another good reason to consult with solutions architects or an equivalent resource as you evaluate data center partners.
Drawbacks of colocation data centers include:
- Capital Investment – While colocation customers experience lower total cost of ownership over time, initial capital investment into infrastructure equipment is required.
- Availability – Currently, vacancy rates are very low, which could delay establishing or expanding a deployment.
- Equipment Management and Maintenance – On the one hand, you have less access to your equipment than if it were on-premises, and once a customer’s IT assets are installed, the configuration, performance and monitoring of their systems are primarily their responsibility.
CoreSite Multi-Tenant Data Centers: Customer Scenarios and Solutions
We often describe CoreSite campuses as interconnection hubs – carrier‑neutral, ecosystem‑rich facilities where enterprises, clouds and networks meet.
The fact is, if your data is moving outside the walls of the enterprise, it is going through colocation data centers, no matter where you locate workloads. Whether privately or over the internet, that traffic frequently traverses across multi‑tenant data centers.
I recently talked with a current customer who was trying to push a lot of data over the internet to a specific cloud provider for storage. Their challenge was that their hardware was outdated, so it couldn’t meet storage requirements, and the lead time for getting additional hardware is very long.
The solution they chose was to connect to a storage service provider located in one of our data centers. All that was required was a cross connect between the customer and the service provider. It’s proof-of-concept of why digital ecosystems, enabled in multi-tenant data centers, matter.
Another customer was also considering using the internet for high volumes of data transfer, unaware that data egress fees would be imposed when they brought data out of the cloud. What I proposed was to use the Open Cloud Exchange® to order and manage the data over our on-site cloud on-ramps, which are direct cloud connections. It gave them control and cut egress fees.
They get approximately 4x cost savings doing that. That’s because it typically costs about eight cents per gig to move data over the public internet, although that number kind of goes down a little bit as you push more traffic. The cost is about two cents per gig by utilizing a CoreSite direct on-ramp to a cloud provider.
That’s just two recent examples. We hear of all kinds of challenges for customers, some pretty common and many new ones. We are seeing a lot of inferencing workloads as AI evolves.
Let Us Architect a Solution for You
I’m a Solutions Architect at CoreSite. My job is simple: listen first, then map out clear, practical options to get you where you want to go. We typically first evaluate the way a customer is operating and then we recommend different approaches. We’ll lay out a few options with the tradeoffs on cost, performance, timeline and risk. Once a customer evaluates the options and decides, we help make it real.
Part of my role also is to be the “glue” between the Sales, Customer Success, Project Management and Operations teams. When I have insight into your infrastructure, I can be a liaison and ensure everyone is clear about your goals and your potential solutions.
For the record, I’m architect‑first, vendor‑second. Sometimes the right answer isn’t the highest‑revenue path—or even a CoreSite component. If there’s a better route, I’ll tell you and help you connect the dots by providing personalized experience for each customer that comes to our team.
So, when you are ready, let’s talk.