Translate this website:
Search this website:

Unified fabric management

By Rupin Mohan, Business & Technology Strategy Manager, Hewlet-Packard, and J Metz, Program Manager FCoE, Cisco Systems.


Date: 9 Jan 2012

“Unified’ and ‘Converged’ may be the most over used terms in technology now days. In this age of austerity, it has become the trend to combine two or more technology functions, be it software or hardware and call it either converged or unified.

Let’s get more specific to data center networking. With the development of Fibre Channel over Ethernet (FCoE), customers have a choice to run IP traffic (LAN) and storage (Fibre Channel SAN) traffic on the same physical wire and switches. If you are using multiple Ethernet and Fibre Channel connections and wires to manage the network and storage traffic in your operating environment, then you are likely concerned about the density and sprawl of switches in your data center and related management and maintenance costs to effectively manage this network. As a potential solution to the sprawl and increased management costs, FCoE as a technology should be on your radar.

As Data Center architectures become more and more capable, the need for comprehensive management tools are more and more pressing. In fact, they’re less of a luxury and more of a requirement. But, the sixty four thousand dollar question is - what qualifies as a “comprehensive management tool?” Vendor implementations take on various metaphors for what a “unified fabric” is versus what “converged fabric” means.

For instance, one view refers to ‘unified’ as the large-scale symbiosis of servers, storage, and network, while ‘converged’ or ‘flex’ refers to the ability to transport different types of LAN and SAN traffic across the same networking infrastructure. Another popular view is exactly the opposite where ‘converged’ refers to servers, storage and network in a box and ‘unified’ or ‘flex’ refers to the ability to transport different types of LAN and SAN traffic across the same networking infrastructure. Still others may have their own mechanisms for defining terminology.

Rather than choose sides or be proscriptive, this article examines the criteria that should be used to determine the appropriate tools that can help you identify what unification management looks like.

There are several other dimensions to this story. Let’s discuss three:
1. Rise of solutions comprising of server, network (IP and/or FC) and storage in a box. These are sometimes tightly and sometimes loosely integrated. But, overall, the big value proposition is that these solutions are fully qualified, easy to buy, easy to deploy and easy to service from a customer point of view. This is what the industry is generally referring to as ‘Converged Infrastructure’ or “Unified Computing’. A flexible network / fabric is one of the key architectural tenets of this solution.
2. Server virtualization gaining significant traction in the market. When customers think of Cloud computing, the first thing that comes to their mind is server virtualization. If you think about it, server virtualization serves as the prime example of unification / convergence in today’s data center. In this age of more of ‘more for less’, convergence from server virtualization delivers tangible results. Note that multi-core CPU’s, dramatic increases in server I/O bandwidth, and high speed network interconnects like 8Gb Fibre Channel and 10GbE DCB Ethernet have been the key foundational drivers of server virtualization.
3. Management Software becoming a control point or sticking point in the data center. The rise of management software from server virtualization vendors as a ‘unified’ management tool which manages your server, network and storage assets is a reality. There is also software from switching providers attempting to manage unified storage area networks and Ethernet networks. This is an area again where there are products that are loosely coupled but strategically, vendors are talking about a single software management platform that manages the ‘unified’ or ‘converged’ network in the data center.

Let’s dig a little deeper into management software. Every storage and storage network vendor provides some level of management for its products. The issue really begins and ends with the concept of scale, and what happens when many pieces of equipment are connected together – especially in a heterogeneous environment.

To that end, the 5 basic things you need to look out for are:

1. Comprehensive, ground-up approach to converged/unified network managementAs most administrators are painfully aware, network and SAN equipment providers historically have released loosely coupled management tools primarily to be able to check that box in the RFP’s. This often leads to having different management parameters and requirements across the same vendor portfolio. This can pose some issues, particularly in the area of unintended consequences between the interconnection and interrelationship between pieces of equipment.

Understanding the traffic within a network takes more than just managing at a device-level. When it comes to storage it is critically important to understand from the “ground up,” from the frame’s initiation point at the initiator, all the way through to its target. Requirements for IP traffic are different compared to requirements of storage traffic in terms of latencies, packet drop ratios, timeouts, etc. This means being able to track the elements where frames may run into potential problems becomes more important. Understanding where problems may occur and make it visibly easy to identify at a glance become extremely useful in preventing issues before they become critical.

Network management tools should allow you the ability to use pre-existing ‘canned’ templates or customized configurations that can help rapid deployment of new equipment with pre-set settings. Wizard-based functionality may also be available for automating mundane tasks that may be error-prone (e.g., such as device zoning, converged LAN/SAN port configuration, etc.).

2. Single versus heterogeneous supportThere is another interesting angle that could be important to data center administers and that is if this software manages products from a single vendor versus products from heterogeneous vendors. Customers are realizing the value of competition and dual sourcing when it comes to purchasing IT assets and also of buying best in class products. Hence, the importance of heterogeneous support is increasing and any management software that can manage all these assets from a single pane of glass does add considerable value to the administrator.

3. Role-based access controlsWhether you call the environment a “unified fabric” or “converged network,” one thing remains the same: eventually there will be multiple teams of people interacting with the same equipment. From a storage perspective, the obvious questions arise: How do you prevent someone from messing up your storage, whether intentionally or by accident?

Without question the ability to set access control limits based upon roles is an unqualified “must.” Any time there is the potential for conflict when using ubiquitous equipment there must be a means to mitigate it within software, regardless of how the network/storage organizational structure works within your company, the management tool should provide this functionality.

4. Threshold alerts and trending analysis/reportsPerformance monitoring is a key element to end-to-end network management, whether it be for LAN or SAN environments. Management software should – at a minimum – be able to identify the current and existing traffic patterns from source to destination for all types of traffic, and be able to easily and quickly distinguish between them.

In addition, there should be an ability to track and analyze network performance over time. By being able to run traffic prediction analyses these tools can often enable the administrator to run “what if” scenarios within the data center, including workflow mobility impacts and virtualization loads on storage arrays, not to mention the impact of adding new applications to the network load.

Logging and reporting becomes critical in these cases. Some organizations wish to begin chargeback to departments for these services, particularly as companies begin to examine the notion of “private clouds” with Infrastructure-as-a-Service (IaaS).

5. 3rd Party API IntegrationAnother interesting development in the management software area is the rising importance and interest in developing 3’rd party API’s to the core management modules. The core idea behind this is similar to the rise of applications in smart phones. Vendors are expecting customers and OEMs to develop custom applications using these 3’rd party API’s which solve individual problems and add specific value. Whether it be a storage tool that must connect with a server virtualization tool, or a networking tool that needs to have interactivity with servers, management tools must at least be able to call on each others’ APIs to be able to poll and process critical data. These API’s further contribute to the ‘network effect’ in the data center network wherein customers get hooked on to features and functionality from a particular vendor.


As you can see, this is an area of enormous innovation and intense competition where vendors are working hard to differentiate and create value for customers. The taxonomy is evolving and requirements for unified / converged management software are being developing. One thing is certain, that management software would be a significant control point for in the data center for converged / unified solutions.

« Previous article

Next article »



Latest IT jobs from leading companies.


Click here for full listings»