Data Center

A Hybrid Cloud Connectivity Template for Colocation Providers

by asardell ‎06-01-2017 07:15 AM - edited ‎08-17-2017 09:17 PM (9,210 Views)

In the last couple of months, we’ve explored how colocation facilities (Colo), especially those that provide Internet Exchange (IXP), can enhance their service offerings for hybrid cloud connectivity.

 

Most recently, we took a close look at enterprise application requirements to provide a high-level menu of where they might reside in terms of a private or public cloud.

 

Here, we’ll explore:

 

  • Prerequisites for creating enterprise hybrid clouds that include multiple Cloud Service Providers (CSP)
  • Connection options both inside and outside the Colo
  • Potential service offerings that bring the enterprises the CSPs together
  • Customer experience requirements
  • A use case showing how you can “broker” a sophisticated multicloud service

This allows colocation service providers to be a viable option in a cloud-centric world, meeting the needs of cloud-first, Mode 2 enterprises, as outlined in the Gartner report entitled, Four Strategies for Colocation Technology Product Management. And as a service provider, it allows you to differentiate yourselves effectively from your competition, as outlined in another Gartner report: Market Insight: Service Provider Differentiation (Part 2).

 

Prerequisites for Hybrid Cloud Connectivity

 

Hybrid cloud connectivity attributes can be defined both in terms of the underlying infrastructure and the automation system.

 

In terms of network connectivity, the requirements depend on the application, but the key attributes are:

 

  • Security: Ensure secure data transfers in and out of the public cloud
  • Latency: Ensure fixed/known latency for connecting to public cloud
  • Availability: Multiple routes to public cloud destinations
  • Performance: Measurable and fixed network bandwidths

All of this can be controlled through the use of visibility services at Layers 2-4; payloads, of course, are off limits for privacy reasons. But nearly all applications can be handled at this level:

 

  • High bandwidth applications such as disaster recovery, big data and analytics, and content storage almost never have real time requirements
  • Some low bandwidth applications (voice, CRM) have near-real time needs, while others (email, reporting, HR or expense) are asynchronous; it is often relatively straightforward to sort them at Layer 4

 

When tuned to the proper degree to match the enterprise and CSP requirements, you can be sure to meet service level agreements (SLA) for all parties.

 

Of course, you also need an advanced automation system to be able to handle these requirements at scale; this system should be able to evolve into one that includes:

 

  • Service Oriented Automation: including one-click connectivity and the ability to consume, and quickly modify, customer configurations
  • Intercloud Service Orchestration: the cross-domain ability to orchestrate services among multiple cloud service providers

Connection Options

 

There are multiple ways you can go, and the choice largely depends on the preferences of your enterprise and CSP customers, as well as what you’re geared up to support inside the Colo.

 

The simplest (most primitive) option is just to connect over the Internet via a network service provider to the CSP. This has the following pros and cons:

 

  • Pros: Can use existing internet connection to enable hybrid cloud; it’s easy and you can connect to any number of CSPs
  • Cons: Security is a huge issue here, and latency can’t be controlled

 

So this option won’t work for anything that requires a decent SLA.

 

An intermediate choice is to support a Point of Presence (PoP) for enterprises and CSPs within the Colo. This brings with it some new advantages:

 

  • Pros: The fixed latency and guaranteed bandwidth of a private LAN
  • Cons: Negotiating with the enterprise to house a demarc in your data center, and the connectivity between the enterprise and the Colo is still over an SP network

Probably the most flexible option is to remove the requirement to connect over the SP network and house the enterprise DC in your facility (Figure 1).

 

 

Ent DC in Colo Private Net1.png

Figure 1: Enterprise DC Housed in the Colo for Maximum Flexibility

 

Of course, this isn’t always possible, and many enterprises will keep much of their infrastructure on premises, but when you can achieve it you have the ability to offer very sophisticated services.

 

Potential Service Offerings

 

Figure 2 shows a high level overview of the service categories you can offer.

 

High Level Service Cats2.png

Figure 2: High Level Service Categories

 

These categories can be described as follows:

 

  • Network to Cloud: Interconnection over a public or private network (WAN or DC) to a cloud service
  • Cloud Exchange: Enable enterprises to connect to multiple cloud providers on a private network
  • Intercloud Service Orchestration: enterprise and/or colocation facility can access and chain network and cloud services

You thus have the makings of a tiered service offering, with the latter two options providing the most flexibility and performance. The first two of these you can do today, and with judicious use of workflow-based automation, you can set up very effective hybrid cloud automation.

 

The final category is more forward-looking, and requires an advanced automation platform to be able to perform cross-cloud orchestration. Even before the full bloom of this vision is realized, elements of solutions in this area can be deployed in an incremental fashion even today. 

 

Optimizing Customer Experience

 

The options for cloud exchange and intercloud service orchestration provide ways to optimize the customer experience for all tenants: enterprise, CSP, and at times even network service providers may house equipment in the Colo.

 

For enterprise tenants, the advantages are:

 

  • One stop for multiple cloud services
  • Connection options to match the application SLA
  • Flexibility to update the connection as the requirements evolve for either hybrid or multicloud

CSP tenants also reap benefits:

 

  • Brokerage between multiple cloud services
  • They can provide their services more easily and cost effectively
  • They can gain customers faster and at higher speed

 

These advantages can be ensured by using architectures to support scale and performance. This should include an underlay to offer custom connections (at variable speeds for physical or logical interfaces) for selected cloud services, and perhaps even Bandwidth on Demand (BoD) from a customer portal.

 
If you are running MPLS or VPLS today, then using technology that can help you migrate to BGP-EVPN with VXLAN, especially while coexisting with your existing underlay, will allow you to scale and perform in conjunction with the needs of your customers.

 

Use Case: Brokering and Orchestrating a Service

 

In this example, we show a Colo houses multiple enterprise, network and cloud services. Here (Figure 3), there are three cloud services being used, but the enterprise has subscribed to only one of them: SalesForce.com (SFDC).

 

multi cloud broker1.png

 

Figure 3: A Service Across Multiple Cloud Providers

 

At the end of every quarter, the enterprise needs to generate a financial report for the quarter. In part, the order is fulfilled by the Colo provider with the necessary data to generate it.


The enterprise provides the location of the data and the requisite algorithms to the Colo provider, and the Colo orchestration engine communicates with SFDC’s orchestration to create a bundle of the data that needs to be used for the report.
 
In the next step, Colo orchestration communicates with a Hadoop provider (AWS in this case) and creates a temporary tenancy for “number crunching” this data.  To do this, Colo orchestration moves the data from SFDC to AWS via the Colo network.


The AWS tenant crunches the data and creates the data necessary for generating a financial report. Colo orchestration communicates with a different CSP (GoogleApps in this case) that will create a new tenant to generate a financial report. After the Colo network moves the data from AWS to Google, the AWS tenancy is terminated.
 
Data is then processed in a report generated and provided to the customer, and the Google tenancy is terminated.

 

Call to Action

 

Be sure to contact your Brocade representative for more information.

 

Related Products and Blogs

 

Follow the links in the article; additionally, the follow related products and blogs will give you more information.

 

Products

 

Brocade technologies that are pertinent to this topic include:

 

Colocation-Related Blogs

 

See also the following blogs on the colocation and exchange industries: