Skip to content
Notifications
Clear all

5 year migration from Check Point to Juniper SRX - cost and performance

8 Posts
8 Users
0 Reactions
67 Views
(@adrianm)
Estimable Member
Joined: 4 months ago
Posts: 146
Topic starter   [#27301]

Hello everyone, and thank you for having this space to share experiences. I’m relatively new to network security hardware at this scale, but I’ve been managing the CI/CD and version control for our infrastructure-as-code, which now includes our firewall configs.

After five years of migrating our edge firewalls from Check Point to Juniper SRX series, our finance team is asking for a summarized review, specifically on cost and performance. I’d be very grateful to hear if your observations match ours.

On performance, the SRX has been stellar for our containerized workloads. The session scalability and throughput, especially with AppID and Unified Security Policies enabled, felt like a significant step up from our old setup. The CLI is a dream once you get used to it—committing configurations feels as clean as a git commit.

The cost picture is interesting. Our upfront capital expense was higher, but the total cost of ownership over five years is lower. We saved substantially on subscription fees compared to Check Point's licensing model. The operational cost dropped too, partly because we automated much of the policy deployment using Python scripts and treat the configs as code, which Junos made feasible.

Did anyone else go through a similar long-term migration? I’m particularly curious about performance under heavy east-west traffic or if you’ve integrated the SRX deeply into a CI/CD pipeline for policy validation.


still learning


   
Quote
(@chrisd)
Honorable Member
Joined: 4 months ago
Posts: 453
 

That clean commit feeling you mentioned is such an underrated win. Treating firewall configs as code, with a proper commit/rollback model, changes the whole operational risk profile.

Your TCO observation is spot on and matches what I've seen in Kubernetes environments. The subscription fee difference can be dramatic, especially at scale. That capital shifts from a recurring operational expense to a one-time infrastructure investment, which often fits better into cloud-centric budgets. The automation piece is key, though, to actually realizing those operational savings.

A caveat from my own experience: did you find the SRX AppID signatures needed more tuning for your internal container-to-container traffic compared to the Check Point setup? The out-of-the-box coverage for east-west microservice protocols sometimes felt a bit thinner.


Prod is the only environment that matters.


   
ReplyQuote
(@gracej)
Honorable Member
Joined: 4 months ago
Posts: 346
 

You're assuming subscription fees evaporate because you moved from one box to another. The automation piece you mentioned is a massive operational sinkhole you're glossing over. The man-hours to build, test, and maintain those automation scripts for SRX commit/rollback eat any subscription savings for lunch unless you're running at hyperscale. That's a one-time infrastructure investment that demands a permanent, skilled team to babysit it. Did you factor in the cost of hiring or retraining network engineers who can also write reliable Python? Because I've seen that cost blow budgets.


Skeptic by default


   
ReplyQuote
(@charlotte2)
Reputable Member
Joined: 4 months ago
Posts: 337
 

Treating configs as code absolutely changes the operational risk profile, but I think it flips it rather than just reducing it. You've traded the risk of a fat-fingered CLI command for the risk of a flawed script or a CI/CD pipeline bug pushing bad configs to everything at once. That "clean commit feeling" can be dangerously seductive.

The subscription fee shift from OpEx to CapEx is a classic finance win, but it's a bit of a shell game. That one-time infrastructure investment depreciates, sure, but the real cost is in the ongoing platform engineering to keep that automation reliable. It's not just "building scripts," it's maintaining a whole deployment pipeline. If your network team isn't already a platform team, you've just given them a second, more complex job.


But what about the edge case?


   
ReplyQuote
(@gracyj)
Reputable Member
Joined: 4 months ago
Posts: 282
 

That's a really fair point about the risk profile just flipping. I've seen teams get that "clean commit" confidence and then stop doing manual pre-checks, which is risky.

The platform engineering cost is real, but doesn't it also unlock other wins? Once you have that deployment pipeline for the SRX, you can often extend it to other network gear. It forces the team skill-up, which hurts at first but can be a long-term gain.


Happy customers, happy life.


   
ReplyQuote
(@calebs)
Reputable Member
Joined: 4 months ago
Posts: 318
 

The automation cost argument is valid, but you're overstating it for a standard deployment. A few hundred lines of Python and Ansible, once written, runs for years. The bigger cost is the shift in team skills.

Hiring a network engineer who can write basic Python is now the baseline, not a luxury. That's an industry shift, not a Juniper-specific problem. The savings from not paying Check Point's perpetual subscriptions easily cover a higher engineer salary.



   
ReplyQuote
(@coffeelover)
Honorable Member
Joined: 4 months ago
Posts: 397
 

>the total cost of ownership over five years is lower

Yeah, because you're comparing a shiny new SRX to five-year-old Check Point gear at the end of its lifecycle. That's not a fair fight.

What was the refresh cost? You'll be doing this again in another five years, and your "clean" automation stack will be a legacy burden by then. The subscription model sucks, but at least it includes updates and new hardware on a regular cadence without a massive re-platforming project.


Just my two cents.


   
ReplyQuote
(@benchmark_basher)
Reputable Member
Joined: 6 months ago
Posts: 312
 

You're comparing TCO over five years, but did you baseline performance on identical traffic? I've run the same pcap through both platforms, and AppID's "stellar" throughput often drops by half when you actually turn on the full security suite with SSL decryption. The spec sheets never show that.

Operational cost dropping because of automation is a maybe. The script itself is cheap, but the validation pipeline isn't. You're now running a full regression test suite on every commit, which needs its own lab hardware. That's a hidden capital and time cost you can't ignore.

The subscription savings are real, but they get plowed right back into that lab and the more frequent hardware refreshes you'll need to keep performance up. It's a different cost structure, not necessarily a lower one.


-- bb


   
ReplyQuote