I'm evaluating the Sophos XGS series for a potential edge deployment across several branch offices, and the vendor documentation heavily promotes the integrated web cache as a key feature for WAN optimization. The marketing claims are predictably vague, focusing on "reduced bandwidth consumption" and "improved user experience" without providing tangible, quantifiable metrics.
Before I consider architecting around this feature, I need to see empirical data. Has anyone conducted a controlled analysis to measure the actual bandwidth savings in a production environment? I am particularly interested in:
* **Methodology:** How did you structure the test? Was it A/B testing with cache enabled/disabled on identical traffic profiles, or longitudinal analysis over weeks/months?
* **Traffic Profile:** What is the composition of your outbound web traffic? Percentages for static content (JS, CSS, images, video) versus dynamic API calls would be crucial.
* **Cache Hit Ratio:** What sustained cache hit ratio did you observe? The Sophos CLI should provide this.
```bash
# Example commands to query cache stats
> ssh admin@xgs
XGS> show web-filter cache statistics
```
* **Quantitative Results:** What was the measured reduction in upstream bandwidth? Please provide it as a percentage of total HTTP/HTTPS traffic and, if possible, in absolute Mbps/GB per day.
* **TLS/HTTPS Overhead:** Given that most traffic is now encrypted, how does the cache handle TLS? Is it performing MITM decryption, and if so, how are the compute overhead and certificate management costs factored into the total savings?
Anecdotal claims of "it feels faster" are not actionable for capacity planning. I am looking for data that can be modeled, such as: "With a workforce of 200 generating approximately 150 GB of daily web traffic with a 40% static content ratio, the cache achieved a 65% hit rate, yielding a net WAN bandwidth saving of 39 GB per day."
Without this level of granularity, it's impossible to perform a credible ROI calculation versus alternative strategies like explicit forward proxies or even increased bandwidth commitments.
-- alex
Great question. I'm curious about this too, especially the traffic profile part. If most of your office traffic is to dynamic SaaS apps like O365 or Salesforce, wouldn't the cache do almost nothing? It seems like you'd need heavy repeat visits to static assets to see real savings.
Did you get a chance to check the CLI for those cache stats on a demo unit? I wonder if the hit ratio is even logged clearly for different content types.
CloudNewbie