Blog

Struggling to manage SD-WAN performance on a budget? Metadata could be the answer!

If stream to disk is proving cost-prohibitive for SD-WAN performance management, consider metadata instead. It could be the missing piece in your data collection strategy to avoid ballooning operations costs. 

According to Gartner, 60% of enterprises are expected to deploy SD-WAN by 2024. SD-WAN is appealing for lots of reasons, including alignment with a shift towards cloud applications, visibility into application performance, and flexible bandwidth and routing control. 

But properly troubleshooting SD-WAN performance requires a mix of active (real-time) and passive (historical) monitoring techniques. The passive component means collecting potentially a *lot* of historical performance data to identify trends and troubleshoot issues. Storing and processing all that data gets expensive.

For many IT organizations, that’s a problem because network budgets specifically are roughly flat and are focused on ‘keeping the lights on’ and running the business. 

All of this leaves IT organizations in the tough position of needing to do more with less. Something has to give. 

Luckily, what has to give needn’t be SD-WAN service quality. Instead, IT organizations need to reexamine what and how much historical network performance data is collected, and how it’s used. Careful planning and strategic choices in this area is crucial.

How? Keep reading! We’re about to get technical!

Stream to disk vs metadata for SD-WAN performance management

One often overlooked option to reduce total cost of ownership (TCO) for passive application and network monitoring is leveraging metadata rather than stream to disk. 

‘Metadata’ here means extracting only transactionally-relevant data (packet headers) to understand performance. This provides synthetic, high-level information defining essential characteristics for performance measurements, on one or more aspects of the underlying raw data, without having to transmit the entire flow. This uses significantly less bandwidth and less storage than stream to disk.  

Metadata capture vs. stream to disk for SD-WAN performance monitoring: how the methods work | Accedian
Metadata vs stream to disk: how they work

While the two technologies can provide similar results, their requirements and cost can vary greatly. As noted, stream to disk comes with large storage and CPU requirements. By comparison, metadata consumes 20x to 100x less storage capacity!

Comparing the storage requirements over a 7-day period for stream to disk vs. metadata for SD-WAN performance management | Accedian
Stream to disk versus metadata storage capacity usage for 7 days 

The cost benefits of metadata alone should make network infrastructure leaders take notice! But also consider two functional downsides of relying only on stream to disk:

  1. Default retention time is usually short; increasing this requires more storage, consequently increasing TCO
  2. Not a fit for dynamic or cloud environments; users target specific points for traffic capture and this is hard to do in an environment where endpoints are always changing
Pros and cons of stream to disck vs. metadata capture for SD-WAN network performance management | Accedian

Pros and cons of the two main approaches to passive traffic analysis

There are some applications where stream to disk does make sense, of course, because of its packet level analysis capabilities. Regulation compliance comes to mind, for example. But, it is simply too expensive, and its retention time is too short, to be an effective solution for monitoring SD-WAN quality of experience (QoE). 

Making the most of metadata for SD-WAN QoE management

Clearly, passive application and network monitoring is the primary tool of choice when you want to get the most accurate picture of user experience. But, to make this effective, you must be able to answer two critical questions: 

  1. What data do I need? 
  2. How far back in time do I need to look? 

The answers to those questions will almost inevitably lead you to metadata as the preferred solution, because:

  • Its depth and breath more than satisfy the requirements for SD-WAN performance monitoring down to the transaction level over a period of months rather than days. 
  • It greatly simplifies retrieval, presentation, and correlation of network performance data.
  • Its longer retention time is critical for monitoring moves and changes. 
  • It is a lightweight software solution that provides detailed information but also aligns with limited IT network budgets. 

With metadata, you can add application QoE monitoring to your SD-WAN deployment—thereby achieving visibility to the true edge (the end user’s experience)—without breaking your IT budget!