Skip to content
Notifications
Clear all

Concrete numbers: Delinea increased our patching time by 15%

7 Posts
7 Users
0 Reactions
49 Views
(@hannahd)
Reputable Member
Joined: 4 months ago
Posts: 216
Topic starter   [#25675]

We switched to Delinea Secret Server (cloud) from a legacy on-prem solution about eight months ago. The goal was better security and workflow. The security part is fine, but the operational efficiency took a major hit we didn't anticipate.

Our main issue is with the API and automation for patch deployments. In our old setup, a script could retrieve a secret and inject it into a deployment workflow in under 2 seconds. With Delinea, the same process now takes 8-10 seconds on average. The overhead appears to be in the authentication handshake and subsequent secret retrieval calls, which are slower than documented.

This added latency has a cascading effect:
* Each patch cycle involves hundreds of automated credential retrievals.
* What was a 40-minute automated patching window is now consistently 46+ minutes.
* That's a 15% increase in operational time, which translates directly to longer system exposure and more after-hours work for the team.

We're on their "Enterprise" plan, and support's response was essentially "within SLA." The SLA is for uptime, not performance. We're now looking at having to architect a local caching layer, which adds complexity and a new failure point—something we were trying to avoid by moving to a cloud PAM.

Has anyone else benchmarked their automation workflows pre/post-Delinea? Specifically:
* Average time for a machine-to-machine API call to retrieve a single secret?
* Any configuration tweaks that actually improved retrieval speed?

We're locked in for another year, so I'm looking for practical workarounds, not alternatives.

—hd


—hd


   
Quote
(@brianw)
Reputable Member
Joined: 4 months ago
Posts: 242
 

Your latency numbers are concerning but track with what I've seen when moving from lightweight on-prem APIs to cloud-based vendor services. The authentication overhead you mentioned is often the hidden cost. A 2-second to 8-10-second per-call increase is massive at scale.

Have you quantified the cost impact of that 15% longer patching window? Consider the compute time for your orchestration hosts staying online longer, plus the team's after-hours labor. The operational inefficiency might be financially equivalent to a sizable chunk of the Delinea subscription itself. Their SLA covering only uptime is a common frustration; performance guarantees are rarely included, shifting the risk entirely to the customer.

Architecting a local cache is a valid workaround, but you're right about the complexity. You'll need a TTL policy, a secure storage mechanism for the cached secrets, and monitoring for cache freshness. It effectively rebuilds a piece of the performance you originally paid for.


Spreadsheets or it didn't happen.


   
ReplyQuote
(@docker_diver)
Honorable Member
Joined: 5 months ago
Posts: 496
 

Yeah, the SLA point hits home. Uptime is easy to measure but performance impact is the real killer. If their API is "up" but responding slowly, the SLA is useless for your workflow.

You mentioned a local cache. Would that mean running something like a small Redis container just to hold secrets temporarily during the patching window? Sounds like extra infra to manage, but maybe it's simpler than it seems. Have you tried that before?


Containers are magic, but I want to know how the magic works.


   
ReplyQuote
(@cloud_cost_hawk_new)
Reputable Member
Joined: 7 months ago
Posts: 333
 

Local cache just shifts the cost around. Now you're running and securing a Redis instance, probably on a cloud VM you pay for by the second. That's not a workaround, it's a self-inflicted integration tax to make their slow API usable.

You're right that performance SLA is what matters. The vendors know this, which is why they only commit to uptime. The financial risk of a "working" but sluggish system lands entirely on your ledger in extended compute time and wasted engineer hours.


-- cost first


   
ReplyQuote
(@crm_hopper_alt)
Reputable Member
Joined: 6 months ago
Posts: 357
 

That 15% operational increase is the real cost they never show on the pricing page. The "Enterprise" plan usually means you're paying more for the privilege of getting the same canned support responses.

The caching layer idea is just accepting their failure as your new project. You'll spend weeks building and validating it, and then you own the security and uptime of that cache too. Did support offer any actual performance tuning, like adjusting timeouts or confirming your region's endpoint? Of course not. It's always "within SLA."


been there, migrated that


   
ReplyQuote
(@davids)
Honorable Member
Joined: 4 months ago
Posts: 568
 

You're right about the risk of just shifting the cost and complexity with a local cache. It becomes another system you have to manage, monitor, and secure.

But I think the key question in the original post has been missed: has anyone measured the performance from different regions or network paths to Delinea's cloud? Sometimes the latency isn't the core API, but the route your automation takes. A quick traceroute during a slow call can be revealing.

And on the support point, I find you often have to escalate past frontline support to get an engineer who can talk about timeouts or connection pooling. The "within SLA" response is a conversation ender. You have to ask for a technical resource directly.


Stay curious, stay critical.


   
ReplyQuote
(@garethp)
Estimable Member
Joined: 4 months ago
Posts: 226
 

The "integration tax" concept you're describing is real, but I'd argue it's often a predictable architectural shift rather than a pure vendor failure. Moving from an on-prem, low-latency call to a cloud service inherently adds network hops and a full authentication flow. The failure is in the sales and migration planning that didn't account for this operational physics.

That said, your point about support is the critical one. When they respond with "within SLA," they are technically correct on uptime but missing the serviceability obligation. A competent support engineer should be able to discuss the OAuth token lifetime settings, whether your automation is using the correct client credentials flow, and the potential for connection pooling or adjusting HTTP keep-alives. If they cannot, you've bought a commodity, not an enterprise solution. Escalation is the only path, as user793 noted.


Plan the exit before entry.


   
ReplyQuote