Scalability is an issue that’s coming up more and more frequently as 10G and 40G networks grow in popularity. As networks grow in size, the ability of network analysis solutions to either handle the growing amount of data or to accommodate the growth is telling of its scalability.
Network growth results in more network analysis through increased analytical throughput, scope, data storage, and distributed analysis. As your network grows and you encounter these issues, there are ways to scale your visibility so that you’re not looking for a needle in a 10G haystack.
Architect for Visibility
As always, knowing your network is key. Know what traffic is important to your company. Is it mission critical business applications, like order entry, financials, and CRM? Or is it web-based traffic that’s driving your online retail business? Once you decide what, and how much, of this traffic requires ongoing monitoring and analysis, you’ll know where to look to specifically identify the traffic that you’ll want to capture. Building visibility into your network infrastructure can help both of these practices. Through strategic placement of analysis points, you’ll be able to get instant information to fix problems faster.
Visibility includes both summary level monitoring data and detailed network metrics, including visibility into network packet traffic and even specific packet decodes. Only a packet-based network analysis system, like the Omni Distributed Analysis Platform, provides the complete range of visibility required to monitor and troubleshoot today’s high-speed networks, keeping networks running smoothly and guaranteeing the very best end-user experience.
Though often the fastest link in your infrastructure, the network backbone – the aggregation of all your distribution layer networks – can be an excellent point for monitoring network traffic and capturing network data for more detailed analysis. Depending on your overall network architecture, the network backbone may be a roll-up of just about all of your critical network traffic, especially if traffic is driven through a centralized network operations center (NOC), or if your company is a heavy user of cloud-based or other third party SaaS applications that drive network traffic through your WAN link. Using high speed network monitoring appliances on the network backbone can centralize your network monitoring and analysis, and save money by consolidating network analysis into a single appliance.
The aggregated traffic on the network backbone will typically be high speed, with more and more enterprises migrating to 10G backbones. Packet-based network analysis on the backbone means you’ll be interested in all of the packets, so you will likely need an appliance like WildPackets’ TimeLine network recorder, which captures at rates up to 12Gbps with zero packet loss. Timeline network recorder allows you to store all your data for forensic analysis while continuously capturing network traffic. And if you’re already migrating your backbone to 40G, you can simply add an aggregation tap and a few more TimeLine appliances for a complete 40G solution.
Adding Visibility to Virtual “Blind Spots”
Traditionally, north-south traffic was the most important in network monitoring. However, with the explosive growth virtualization, east-west traffic is becoming more and more important in enterprise networks, and poses a new challenge in network and application performance monitoring. East-west traffic is typically traffic moving within a virtual host or a distributed virtual system. Since much of this traffic resides solely within the virtual environment, and therefore never hits a physical network interface, traditional network monitoring and analysis that is done by tapping into the physical network does not capture this east-west traffic. For example, let’s say the order entry system and the inventory database reside on separate VMs within the same host or distributed system. Communications between the order entry application and the database are east-west traffic. Application performance issues between these systems are “hidden” within the VM. To add visibility, you can either install WildPackets OmniVirtual on one of the VMs to gain visibility into the entire host, or, in the case of larger, distributed virtual systems, the use of a virtual tap is recommended. Virtual taps are sold by many tap vendors, and they provide a physical link that traditional network monitoring appliances can access to expose east-west traffic within the virtual system.
For more information about how WildPackets can help scale your networks, check out our ondemand webcast.