Search all 478 artefacts by title, document ID or content.
Printed on the standard letterhead. Page furniture, margins and repeating table headers come from the same stylesheet the PDF service uses.
Pensieve Labs
The operating system for hospitals
DIS-GL-030
v1.0.0 | 31 July 2026
A percentage without a measurement method is not a commitment. This document publishes the method; the
commitment itself is in SLA-GL-001 and on the Order Form.
DM-1 Dedicated |
DM-2 Shared |
DM-3 Customer Cloud |
DM-4 On-Premise |
|---|---|---|---|
Availability committed in SLA-GL-001 |
Committed | Committed subject to the hospital's project remaining available, funded and within quota | No availability commitment. Pensieve Labs does not control the hardware, the power, the network or the room |
To state how availability is measured, what is excluded, what performance a hospital should expect, how
capacity is managed, and where Pensieve Labs will not make a commitment. It exists so that a hospital
comparing vendors can compare like with like, which is impossible when one vendor quotes a percentage and
does not say what it measured.
| Element | Definition |
|---|---|
| What is measured | Whether an authenticated request to a core Platform function returns a successful response within the timeout, from an external monitoring point |
| Core functions | Patient registration and search; admission, discharge and transfer; clinical record read and write; order entry; result view; medication order and administration; billing and receipting; pharmacy dispensing. The functions a hospital cannot operate without |
| Measurement interval | Continuous synthetic checks at a fixed interval from an external point, plus real request success rate from the load balancer |
| Unit of downtime | A period in which core functions do not return successful responses. Partial degradation affecting a named function is reported as a partial outage against that function, not absorbed into an aggregate |
| Measurement period | Calendar month |
| Who measures | Pensieve Labs. Not independently verified (Section 6.1) |
| Where it is published | The status page and the monthly service report (Section 4) |
What is deliberately not counted as availability: a fast response that returns wrong data, and a
successful login to a system whose ward terminals cannot reach the network. The first is a defect handled
under SLA-GL-001 severities; the second is outside Pensieve Labs's boundary and is Section 2.
Availability commitments in SLA-GL-001 exclude the following. They are listed here rather than only in the
contract, because a hospital should see them before it signs.
| Excluded | Why |
|---|---|
Scheduled maintenance inside the agreed window, notified in advance per DIS-GL-031 |
Agreed in advance and scheduled to the hospital's low-activity period |
| Emergency maintenance to address a Critical vulnerability or an active incident | DIS-GL-017 Section 4. Notified as soon as practicable |
| Failure of the hospital's own network, internet connection, power, workstations or local devices | Outside Pensieve Labs's boundary |
| Failure of a third-party system the hospital connects to: a payment gateway, an insurer portal, a national registry, an analyser | DIS-GL-024. Pensieve queues and retries; it cannot make another party's system available |
| Force majeure | MSA-IN-001 |
In DM-3: unavailability caused by the hospital's own cloud project (suspension, quota, budget, organisation policy) |
The hospital owns the project |
In DM-4: everything. No Pensieve Labs availability commitment applies to hospital hardware |
ADD-GL-008 |
On DM-4, plainly: Pensieve Labs publishes no availability figure for hardware it does not
own, in a room it has never entered, on a network it cannot see. SLA-GL-001 limits the DM-4 commitment
to Pensieve Labs's own response times, not to uptime. A vendor offering the same uptime number for
on-premise hardware as for its own cloud is either not thinking or not telling the truth.
Performance targets are stated as server-side response time at a percentile, because an average hides the experience that actually annoys a nurse: the slowest one in ten.
| Class of interaction | Target | Measured |
|---|---|---|
| Interactive read: open a patient, view a result, load a worklist | [TO BE SUPPLIED] at the 95th percentile |
Server-side, at the load balancer |
| Interactive write: save a note, place an order, record an administration | [TO BE SUPPLIED] at the 95th percentile |
Server-side |
| Search across the record | [TO BE SUPPLIED] at the 95th percentile |
Server-side |
| Report generation and bulk export | Asynchronous. The hospital is notified on completion rather than waiting | Not applicable |
| Integration call to an external system | Bounded by the far end. Pensieve applies a timeout, queues and retries (DIS-GL-024 Section 16.2) |
Recorded per integration |
Stated plainly: as at
31 July 2026Pensieve Labshas not published measured percentile figures, and the targets above are[TO BE SUPPLIED]rather than asserted. Publishing a numberPensieve Labshas not measured under representative load would be exactly the kind of claim this Trust Center exists to avoid. The figures are published with the first monthly service report and re-published monthly thereafter.
What Pensieve Labs will not measure. Client-side rendering time on a hospital's own workstation
depends on that workstation, its browser, its antivirus software and the hospital's network. Pensieve Labs
measures to the edge of its own boundary and says so, rather than quoting an end-to-end number it does not
control. WPR-GL-001 Section 18.2 item 4 is the corresponding hospital responsibility.
| Artefact | Contents | Cadence | Audience |
|---|---|---|---|
| Status page | Current state of the Platform, active incidents with updates, and scheduled maintenance | Live | Public |
| Monthly service report | Measured availability against the commitment; every incident with its duration and cause; performance percentiles; capacity headroom; releases made; open severity-weighted defects | Monthly | Each hospital |
| Service review | The above, discussed, with actions | Per SLA-GL-001 |
Each hospital |
| Post-incident review | Root cause, timeline, clock performance, remediation with owners and dates | Per incident, per DIS-GL-016 Section 5 |
Affected hospitals |
A missed target appears in the report as a missed target. It is not reclassified, and the monthly report is issued whether or not the month went well.
| Element | Position |
|---|---|
| Scaling | The application tier runs on managed container compute that scales with request volume. Peak clinic hours, a discharge rush and a month-end billing run do not require manual intervention |
| Database capacity | Sized per deployment against the hospital's bed count, transaction profile and expected growth, and reviewed at each service review. Headroom is reported monthly |
| Storage | Object storage grows without a provisioning step. Imaging is the dominant growth driver and is forecast separately |
| Growth planning | The Order Form records the expected scale. A material change (a new site, a new service line, a bed expansion) is a Change Order (ADD-GL-015) so that capacity is provisioned before it is needed, not after a complaint |
DM-2 noisy neighbour |
Rate limiting per authenticated principal and per tenant; per-tenant concurrency limits on background jobs, so one hospital's bulk export does not starve another's ward terminal (DIS-GL-032 Section 5) |
DM-4 capacity |
The hospital's. Pensieve Labs specifies the hardware in ADD-GL-008; the hospital buys, installs and maintains it. Under-specified hardware is a DM-4 performance risk Pensieve Labs can advise on and cannot fix |
| Load testing | Performed before a release that materially changes a hot path (DIS-GL-031) |
6.1 Availability is measured and published by Pensieve Labs, not independently verified. The status
page and the monthly report are Edsol Edtech Pvt. Ltd.'s own measurements. A hospital that wants
independent evidence can run its own synthetic checks against its own tenant, and Pensieve Labs
encourages it: an external check the hospital controls is worth more than a vendor's dashboard.
6.2 No measured performance figures published yet. Section 3.
6.3 No availability history yet. Edsol Edtech Pvt. Ltd. has no published operating history to show. A
hospital evaluating Pensieve Labs should weigh that, and WPR-GL-005 Section 6.6 states the same about the
absence of reference customers. Pensieve Labs will not manufacture a figure to fill the gap.
6.4 No 24×7 staffed watch floor. Monitoring and alerting run continuously and route to an on-call
engineer. Response times are in SLA-GL-001; the limitation is WPR-GL-001 Section 17.2.
6.5 No DM-4 availability commitment. Section 2.
6.6 DM-3 availability is conditional. Section 2. Pensieve Labs cannot commit to a figure for a project
whose quotas, budget and organisation policy it does not control.
6.7 Integration availability is bounded by the far end. Pensieve queues and retries; it
cannot make a payment gateway, an insurer portal or a national registry available.
6.8 A shared estate fails together. In DM-2 a platform-wide incident affects every tenant in the
estate simultaneously. DIS-GL-032 Section 8.5.
| Question | Document |
|---|---|
| The availability commitment, severities, response times and service credits | SLA-GL-001 Service Level Agreement |
| Recovery objectives and restore | DIS-GL-014 / DIS-GL-015 |
| Maintenance windows, release notice and rollback | DIS-GL-031 Change Management & Release Disclosure |
| Incident notification and post-incident review | DIS-GL-016 |
| Multi-tenant resource separation | DIS-GL-032 Section 5 |
DM-4 hardware specification and its consequences |
ADD-GL-008 On-Premise Supplement |
| Capacity changes and their commercial treatment | ADD-GL-015 Change Order |
| Version | Date | Author | Summary |
|---|---|---|---|
| 1.0.0 | 31 July 2026 |
Pensieve Labs Engineering |
First published edition. Measurement method and exclusions published. Performance targets and availability history published as unsupplied rather than asserted. DM-4 stated as carrying no availability commitment. |