What is third-party maintenance?
Third-party maintenance — written 3rd party maintenance about as often, and shortened to TPM — is independent hardware support for servers, storage and networking. This guide covers what it is, what it genuinely cannot do, how the contracts are written, what the market’s cost claims are actually worth, and which of the names on your shortlist are now the same company.
Last reviewed 7 October 2026 · every external figure below is sourced at the foot of the page
What third-party maintenance is
When you buy a server, a storage array or a switch, the manufacturer sells you a warranty and then, when the warranty runs out, sells you a support contract to replace it. That second contract is the one third-party maintenance competes with. It has a different name at every vendor: Dell ProSupport, HPE Tech Care, Cisco Smart Net Total Care, IBM ServicePac, Lenovo Premier Support, NetApp SupportEdge, Oracle Premier Support for Systems, Juniper Care.
A third-party maintenance agreement buys the same hardware outcome from an independent company: someone answers the call, diagnoses the fault, brings the part and puts the machine back into service, against an agreed response time. The spelling varies more than the meaning — third party maintenance, 3rd party maintenance and TPMare the same thing, and UK buyers use all three interchangeably. You will also see “independent hardware maintenance” and “post-warranty support”, which mean the same.
Two situations drive almost every enquiry. The first is a renewal quote that has gone up for equipment that has not changed. The second is an end-of-service-life notice: the manufacturer has decided to stop supporting a platform that is still doing its job, and the choice is to replace it on the vendor’s timetable or to support it another way. Third-party maintenance exists for both, and it is a poor answer to some other problems — which is the subject of the last section.
The four kinds of provider, and why the difference matters
“Third-party maintainer” describes several quite different businesses. In its 2019 Market Guide for this category — the second edition, following a first guide of the same title published 9 August 2017 — Gartner separated providers into four types: pure-play TPMs, whose whole business is maintenance; secondary-hardware TPMs, which also trade in used equipment; TPM aggregators, which assemble cover from other providers; and TPM selling agents, including resellers, which sell someone else’s maintenance.
The distinction is not academic, and it is the single most useful question a buyer can ask. It decides who actually turns up, who holds the spare part, who carries the engineering knowledge when a fault is not obvious, and how many companies sit between your incident and the person who can fix it. Gartner’s own guidance to buyers is to establish whether a provider performs third-line engineering itself or passes it on.
Most of the market is quiet about this, which is why the question works. A provider that delivers its own third-line support will say so without being pushed. A provider that brokers it can still give you an excellent outcome — plenty do, particularly on multi-vendor estates — but you should know which you are buying, because it changes who is accountable when a part is not where it was promised. Ask it of every provider you shortlist, including us.
Who actually owns whom in the UK market
If you are drawing up a shortlist, it is worth knowing that the best-known names in this category now sit inside a small number of groups. The consolidation is a matter of public record:
| Date announced | What happened |
|---|---|
| 11 November 2020 | Park Place Technologies announced the acquisition of Curvature, describing the result as the largest global provider of third-party data centre maintenance. |
| 1 February 2021 | Service Express acquired Blue Chip, then one of the larger UK third-party maintenance and managed-services firms, in its first international deal. |
| 28 June 2022 | Service Express completed the integration of its UK businesses, retiring the Blue Chip and ICC Group brands. |
| 18 November 2024 | Evernex, the other large multi-vendor group active in UK third-party maintenance, acquired Ultra Support, a channel-only UK pure-play maintainer. |
| 6 January 2026 | Park Place Technologies completed a merger with Service Express. The combined company reported more than 3,000 employees and over 20,000 customers. |
The practical effect is that Park Place, Curvature and Service Express are one corporate group, with the former Blue Chip and ICC Group businesses inside it — both UK brands were retired when Service Express completed their integration in June 2022. Three consequences follow, and all three are things a buyer would otherwise find out late.
The first is that a shortlist of those names is not a competitive shortlist. The second is about evidence, and it caught us out too: that group now runs the single most-used public end-of-service-life tracker. Service Express’s separate EOSL database no longer exists as an independent reference — its address now resolves to Park Place’s EOSL library — so two sources UK buyers used to check against each other have become one. We have corrected our own server and network EOSL checker to say so where it relies on it. The third is that when competing contracts are consolidated onto one platform, the terms you renewed under are not necessarily the terms you renew into — so re-read the contract rather than the quote.
What a third-party contract covers, line by line
Contracts vary, but a hardware maintenance agreement in this market is built from five components. Read a quote against these five and the gaps become obvious.
1. Fault logging and triage
A route to raise an incident and get it triaged, in and out of hours. Find out what that route actually is — portal, phone, email — and what happens at 2am on a Sunday, because that is when it matters and it is the detail most often left vague.
2. Diagnosis
Establishing what has failed. On a drive this is quick; on an intermittent memory or backplane fault it is the whole job, and it is where the difference between first-line scripts and genuine third-line engineering shows up.
3. Spare parts
The replacement component, and the logistics that get it to your site. Ask where parts are held relative to your sites, how they are tested before dispatch, and whether they are original manufacturer parts, manufacturer-refurbished or equivalent — the answer is legitimate either way, but it should be in writing.
4. Engineering attendance
Someone to fit the part, whether an engineer attends or you fit it yourself under guidance. For estates with on-site staff, a parts-only contract is often both cheaper and faster.
5. The service level
The time commitments attached to the four components above, which is the next section.
Firmware and software: the real limit, vendor by vendor
This is the one genuine limitation of third-party maintenance, and the market is strikingly quiet about it. A third-party contract cannot grant you a manufacturer’s software or firmware entitlement, because that entitlement lives in the manufacturer’s licence terms, not in your maintenance agreement. Any provider implying otherwise is misleading you.
What matters commercially is not the principle but the detail, and the detail is different at every vendor. Two worked examples, because they are the two that come up most:
Cisco
Cisco publishes its security advisories publicly, and where an advisory offers a free software update — typically for Critical- and High-severity vulnerabilities — a customer with no service contract can normally obtain it by contacting Cisco TAC with the product serial number and the advisory’s URL. Cisco describes this as a customer-service decision made case by case rather than a contractual right, so treat it as a good likelihood, not a guarantee. Feature releases and general bug-fix updates need a current, active licence.
So for Cisco hardware still inside its support window, the security position after leaving SmartNet is better than most buyers assume. The limit matters more than the rule, though, because it bites on exactly the equipment third-party maintenance is usually bought for: once a product passes its Last Day of Support, Cisco states that its PSIRT will continue to accept vulnerability reports but will not analyse, fix or disclose them. Past that date there is no advisory, so there is no free update to ask for — and that, not the maintenance contract, is the real decision point. The detail is on our Cisco third-party maintenance page.
IBM Power
IBM is the opposite case and the one to plan around carefully. On POWER8 and later systems, applying new system firmware requires a valid Update Access Key, and IBM issues those only against warranty or an IBM hardware maintenance agreement. Firmware released while your key was valid stays available to you, and IBM builds its firmware so that HIPER fixes can still be downloaded with an expired key — but anything newer cannot be applied. The practical answer is to apply the latest firmware before the transition, not after. See IBM third-party maintenance.
The general rule that falls out of this: third-party maintenance is an excellent answer for hardware reliability and a poor answer for a platform that needs continuing security patches to stay compliant. If a system carries un-patchable compliance risk, a controlled refresh is the honest recommendation, and any provider worth using will tell you that rather than sell you a contract.
How the SLA notation works, and the question nobody answers
Service levels in this market are written in a shorthand that is rarely explained, and the order of the numbers is not consistent between suppliers. The manufacturers’ own documents read it as coverage hours × coverage days × response target — Cisco’s 8x5x4 is a four-hour response inside an eight-hour business-day window — while many providers and resellers write the same window the other way round, as 5x9 or 7x24. Only the last number is reliable: the response target is always last. Because the first two are genuinely ambiguous, have the coverage window written out in words in the contract rather than trusting the label.
| Notation | Coverage window | Response target | Typically bought for |
|---|---|---|---|
| 24x7x4 | Every day, around the clock | Four hours | Production workloads where downtime is costed in hours |
| 24x7xNBD | Every day, around the clock | Next business day | Resilient kit that still needs out-of-hours fault logging |
| 9x5x4 (also written 5x9x4) | Business days, a nine-hour window; Cisco’s equivalent 8x5 band is eight hours | Four hours | Office-hours systems with staff on site |
| 9x5xNBD (also 5x9xNBD, 8x5xNBD at Cisco) | Business days, a nine-hour window | Next business day | Resilient, clustered or low-criticality equipment |
Now the part that decides whether any of it means anything. A four-hour target is only as good as the answers to three questions, and we would encourage you to put them to every provider in writing.
What event starts the clock?This is the one that varies most and is disclosed least. A clock that starts when you log the call is a very different commitment from one that starts when the provider has completed its diagnosis, and both are sold as “four-hour”. Phrases such as “on site within four hours of diagnosis” or “engineer on site (if needed) in four hours” are doing real work; ask what the trigger is and get the answer in the contract.
What is the target measuring? A response, a diagnosis, parts on site, or hardware restored to service are four different measures, and a contract that names only one of them has not told you much. And when does it stop? Ask what pauses the clock — waiting on your access, your change window, your decision — because a stop-clock rule exists in most agreements whether or not it is published.
EOL, end-of-sale and EOSL: what the dates actually mean
Most third-party maintenance decisions are triggered by a lifecycle date, and the terminology is used inconsistently by the vendors themselves. Three dates matter:
End-of-sale is the last date you can order the product new. End-of-lifeis usually the announcement that starts the withdrawal process, which is why an “EOL” product is often still fully supported. End-of-service-life — Cisco calls it the Last Date of Support — is the date after which the manufacturer will not support the equipment at all. That is the date that forces a decision.
The complication is that vendors publish these very differently, and some do not publish them at all. Cisco is the most transparent: it issues a numbered end-of-life bulletin per product with an explicit Last Date of Support. Lenovo publishes End of Service dates by machine type and model. HPE runs a formal process and publishes both its policy and individual product end-of-life notices openly, but there is no single public list of HPE end-of-support-life dates: for a specific server the date comes from a serial-number entitlement lookup. Dell publishes no model-wide end-of-service-life date at allfor PowerEdge — support is tied to each unit’s service tag, so the only correct answer for a specific Dell server is a service-tag lookup, and anyone quoting you a model-wide Dell EOSL date has got it from somewhere else.
That gap is what the public EOSL trackers fill, and it is why the ownership question in the consolidation section above matters: a tracker date is a secondary source, and two trackers inside one corporate group are not two sources. Our own checkers show the source for every date and say plainly where no re-verifiable date exists rather than publishing a guess — servers and network, storage, and more detail in when OEM support actually ends.
What it costs, and why the percentages you will read are unreliable
Search this category and you will be told third-party maintenance saves 30%, 40%, 50%, 70%, 80% and 90%. Those figures appear on provider websites almost entirely without citations, and they are not different estimates of the same quantity — they are measured against different baselines, which is why they cannot be reconciled.
Only two analyst documents in this category can be named, and both are old. Gartner’s Market Guide for Data Center and Network Third-Party Hardware Maintenance (29 August 2019, ID G00414695) is the one the market quotes, and the band attributed to it is measured against OEM net prices. Gartner’s earlier Competitive Landscape (7 July 2017, ID G00327262) quoted an average of about 60% off OEM list prices, with a range running to 95% off list.
That difference in baseline — net in one document, list in the other — is most of why the percentages in this market cannot be reconciled with each other. If your own comparison starts from a list price, a net-based band will not reproduce, and the reverse is equally true. It is also worth saying plainly that the 2019 guide is the second edition of two and we could find nothing later: the entire market, ourselves included, is anchored to documents that are seven and nine years old, which is more honest than implying a current analyst endorsement.
What this means in practice is that the only number worth anything is the one calculated against your actual renewal quote, device by device, because the saving depends on the vendor, the age of the equipment, the service level and what you are paying now. Some devices will not move at all — anything still in warranty, or needing active software entitlement. A credible quote tells you which of your devices those are.
If you want an indicative figure before you talk to anyone, our third-party maintenance savings calculator applies the published band to your own spend and shows its workings, and the coverage builder works device by device. Both are estimates and labelled as such.
Ten questions to ask any provider before you sign
These are the questions that separate providers, drawn from the gaps that recur across this market. They are deliberately ones that a good provider can answer immediately.
- Do you perform third-line engineering yourselves, on which platforms, and where it is brokered, to whom?
- What event starts the SLA clock, and what pauses it?
- Where are the spare parts for my models held, and how are they tested before dispatch?
- Are parts original, manufacturer-refurbished or equivalent — and will you state that per device class?
- Confirm in writing what happens to firmware and software entitlement on each of my vendors.
- How do I raise a fault at 2am, and what is the escalation path when the first response does not resolve it?
- Can I remove a device mid-term, what notice is required, and is the unused amount credited?
- Which of my devices will you not take on, and why?
- Which accreditations do you actually hold, with certificate numbers and expiry dates?
- Can I speak to a customer running similar equipment on a similar service level?
A provider that answers all ten plainly is demonstrating something more useful than any savings percentage. If any answer is a brochure, ask again.
Third-party maintenance and UK public procurement
For UK public-sector buyers, the route matters as much as the service. The relevant central agreement for technology products and the services attached to them is RM6098, Technology Products and Associated Services 2 (TePAS 2), which runs to 9 October 2027. It is now owned by the Government Commercial Agency, which Crown Commercial Service became on 1 April 2026; existing contracts and call-offs are unaffected by that change. It is structured in eight lots, including hardware, software, and a Sustainability and Circular IT lot that covers refurbished equipment — the lot most relevant to extending the life of an existing estate.
One point worth testing rather than assuming: G-Cloud is a cloud-services framework, and the buyer guidance points to RM6098 when what you need is hardware. Suppliers do nonetheless list hardware maintenance and break-fix services on the Digital Marketplace under its support lot, and a search will return a few hundred of them. So the question is not whether you can find them there — you can — but whether buying physical-asset break-fix through a cloud framework is within scope for your requirement. That is a question for your commercial team to settle before you rely on it, not a detail to discover at contract award.
Two practical notes. Hardware maintenance for public bodies routinely touches systems holding personal data, which brings supplier security requirements into scope, so expect to be asked for certification evidence and check it is current rather than historic. And plan for the framework transition: the successor, RM6398 (TePAS 3), is already in procurement with award expected around August 2027, roughly two months before TePAS 2 closes, so leave margin if you are timing a procurement around it. A call-off signed under TePAS 2 may validly run past that expiry — call-off length is not capped and call-offs survive the framework closing — but it is worth settling at signature which terms govern the contract for its whole life.
When third-party maintenance is the wrong answer
We sell this service, and it is still the wrong choice in several common situations. A provider that cannot tell you which ones is not giving you advice.
When the platform needs continuing security patches. This is the big one. If a system must be patched to stay compliant and its vendor will not supply patches without an active entitlement, extending the hardware does not solve the problem — it defers it while the risk grows.
When the equipment is genuinely too slow or too power-hungry. Maintenance keeps hardware running; it does not make it faster or cheaper to run. Where the current platform is the constraint, the maintenance saving is a rounding error against what a refresh returns — and on power alone, the newer generation often pays for itself.
When parts for the model have genuinely dried up. There is a point past which a spares market stops existing for a specific platform. An honest provider tells you that before the contract, not at the first failure.
When the estate is small enough that the overhead exceeds the saving. A handful of devices may be better served by a spares pool and a time-and-materials arrangement than by a formal contract.
If you are weighing extension against replacement, the comparison with OEM support and the refresh planner both set out the trade-off, and our new versus refurbished calculator covers the middle path.
Where to go next
This guide is the general case. The specifics differ enough by manufacturer and by equipment type that they have their own pages:
Frequently asked questions
What is third-party maintenance?
Third-party maintenance — also written 3rd party maintenance, and shortened to TPM — is hardware break-fix support for servers, storage and networking bought from an independent provider instead of the manufacturer. It replaces the OEM’s post-warranty programme, such as Dell ProSupport, HPE Tech Care or Cisco Smart Net Total Care. It covers diagnosis, engineers, spare parts and a response SLA. It does not supply the manufacturer’s software or firmware.
Is third-party maintenance legal, and does it void anything?
It is a normal commercial service and buying it does not forfeit anything you already own. What it cannot do is transfer a manufacturer’s software or firmware entitlement, because that entitlement is granted under the manufacturer’s own licence terms rather than under your maintenance contract. The practical consequence differs by vendor and is set out in the firmware section above.
How much does third-party maintenance save?
Less reliably than the market’s marketing suggests. The only document in this category that can be named is Gartner’s Market Guide for Data Center and Network Third-Party Hardware Maintenance (29 August 2019, ID G00414695), and the band usually attributed to it is measured against OEM net prices rather than list, while Gartner’s earlier 2017 Competitive Landscape quoted around 60% off list. Figures of 30%, 80% and 90% circulate with no traceable source at all. Treat any percentage without a dated citation as marketing.
Can third-party maintenance support hardware past end-of-service-life?
Yes, and this is its main use. End-of-service-life is a date the manufacturer chooses, not a statement about the equipment. Independent providers maintain spares pools and engineering skills past that date, commonly for a further three to five years and sometimes longer. The limit is usually parts availability for the specific model, not the date itself.
What is the difference between EOL, end-of-sale and EOSL?
End-of-sale is the last date you can order the product. End-of-life is usually the announcement that begins the withdrawal process. End-of-service-life, or last date of support, is the date after which the manufacturer will no longer support it. Vendors use these terms inconsistently and some publish no model-wide date at all, which is covered in the lifecycle section above.
Does one third-party contract cover servers, storage and network together?
Normally yes, and it is the most common reason organisations move. A multi-vendor agreement gives one contract, one escalation route and one renewal date across equipment from several manufacturers, instead of a separate OEM programme and renewal cycle for each.
Sources
Every external date and figure above resolves to one of these. Where no re-verifiable source exists for a claim, the claim is not on this page.
- Gartner, Market Guide for Data Center and Network Third-Party Hardware Maintenance, 29 August 2019 (ID G00414695) — Gartner client-only research; the abstract is public, the full document is not
- Gartner, Competitive Landscape: Partnering With Third-Party Maintenance Providers…, 7 July 2017 (ID G00327262) — the earlier, list-price-based figures
- Park Place Technologies — acquisition of Curvature, announced 11 November 2020
- Service Express — acquisition of UK-based Blue Chip, announced 1 February 2021 (original page withdrawn after the Park Place merger; Internet Archive capture)
- Park Place Technologies — merger with Service Express completed 6 January 2026
- Crown Commercial Service RM6098, Technology Products and Associated Services 2 (TePAS2)
Vendor firmware and software-entitlement positions are summarised from each manufacturer’s own published support and licensing material, cited in full on the per-vendor pages linked above. Last reviewed 7 October 2026. If you find something here that is out of date or wrong, tell us and we will correct it — our editorial standards and corrections policy.
Talk to a UK specialist
Get expert advice or a no-obligation quote — servers, storage, networking, maintenance, finance and cloud. We reply the same working day.