If you're setting up IT infrastructure for a hospital, diagnostic center, or radiology chain in India, the real question isn't "which server should I buy" — it's "which two servers should I buy, and how big." A PACS server and an HIS server do completely different jobs and get sized completely differently; put both on one under-specced box and you'll hit a wall within a year. Here's how to size each one properly, what redundancy actually means for a healthcare setup, and where ABDM/NABH compliance changes your hardware decisions.

PACS vs HIS — Two Different Servers, Two Different Jobs

PACS (Picture Archiving and Communication System) stores and moves medical images — CT, MRI, X-ray, ultrasound, mammography. Its job is almost entirely about storage capacity and read/write speed: a radiologist pulling up a 300-slice CT scan on a reporting workstation needs that data fast, and the server needs to hold years of scans without running out of room.

HIS (Hospital Information System), often bundled with EMR/EHR, handles patient records, admissions, billing, pharmacy, and lab workflows. Its job is about concurrent users and database transactions — dozens or hundreds of front-desk staff, nurses, and doctors hitting the same database at once, all day.

Running both on one under-specced box is the single most common mistake we see in smaller hospitals and diagnostic centers. A busy morning of OPD billing shouldn't slow down a radiologist opening a scan, and a large imaging archive shouldn't eat into the RAM your HIS database needs for query performance. Separate servers — or at minimum separate storage volumes with dedicated RAID — solve this cleanly.

How Much Storage Does a PACS Server Actually Need?

This is the number every diagnostic center owner asks and rarely gets a straight answer to. Here's the real math, based on average DICOM study sizes per modality:

A CT-heavy diagnostic center doing 40–50 scans a day generates 6–8 GB of raw imaging data daily — over 2TB a year, before adding MRI, ultrasound, and X-ray volumes. A 150–200 bed hospital with a full radiology department (CT, MRI, X-ray, USG) running 150–250 studies a day typically needs to plan for 8–15TB of usable storage in year one, growing every year after.

Most facilities under-budget storage by 40–50% in year one because they size for today's volume, not year-three volume. Buy the storage chassis with room to add drives later, even if you don't populate every bay on day one — far cheaper than a forklift upgrade in 18 months.

HIS Server Specs — Sizing by Bed Count and OPD Volume

Unlike PACS, an HIS server's bottleneck is almost never raw storage — it's CPU cores, RAM, and concurrent database connections. Sizing scales with bed count and daily OPD footfall, not data volume:

• Small clinic / nursing home (<50 beds): single server, 16–32GB RAM, 8-core CPU, RAID-1/10 on SSD — handles registration, billing, and basic EMR for 20–40 concurrent users comfortably.
• Mid-size hospital (50–200 beds): 64–128GB RAM, dual-CPU (16+ cores), RAID-10 SSD/NVMe for the database tier — this is where most Indian hospitals sit, and where a refurbished Dell PowerEdge R740 or HPE ProLiant DL380 Gen10 with dual Xeon Scalable processors handles the load with headroom.
• Large hospital (200+ beds) / multi-specialty: clustered setup — a primary HIS/EMR database server plus a standby/failover node, 128GB+ RAM per node, dedicated storage server for lab/pharmacy integration.

A rule of thumb we give hospital IT teams: budget roughly 1GB of RAM per 3–4 concurrent active users on the HIS, then add 30–40% headroom for month-end billing spikes and report generation — that's exactly when under-provisioned HIS servers start timing out.

Redundancy and Uptime — Why Hospital Servers Can't Go Down

A retail business can absorb 20 minutes of server downtime. A hospital can't — a PACS outage means a radiologist can't read a scan mid-emergency, and an HIS outage means the front desk can't admit patients or pull drug allergy history. This is where healthcare IT genuinely differs from every other vertical we build servers for.

Non-negotiables for hospital/diagnostic center servers:

• RAID-10 (not RAID-5) for anything transactional — HIS databases and recent-study PACS storage. RAID-5's rebuild-time risk during a drive failure isn't something you want on a live patient database.
• Dual power supplies, ideally on separate UPS/circuits — a single PSU failure should never take the server down.
• Hot-swap drives so a failed disk gets replaced without a shutdown.
• A tested backup/DR plan — RAID protects against a drive failure, not against ransomware, fire, or a bad firmware update. See our on-premise backup server guide for how to size that separately.
• A real AMC with guaranteed response time. A lot of hospitals get burned here — they buy a server, skip the AMC to save cost, then wait three days for a part when a drive fails. This is exactly the gap Serverwale's server AMC plans are built to close — tested spare inventory and defined response SLAs, because "we'll order the part" isn't good enough when a hospital server is down.

ABDM & NABH — What Compliance Actually Means for Your Hardware

If you're building or upgrading hospital IT in 2026, ABDM (Ayushman Bharat Digital Mission) shapes real infrastructure decisions, not just paperwork:

• Data format standards: ABDM mandates HL7 FHIR R4 for clinical record exchange, DICOM for imaging (which your PACS already speaks natively), SNOMED CT/ICD-10-11 for clinical terms, and LOINC for lab results. The hardware requirement underneath is simply enough CPU/RAM headroom to run the FHIR API layer alongside normal HIS load without degrading either.
• No single-location mandate: ABDM does not require all records to sit in one system — distributed architecture (on-prem HIS + on-prem PACS + cloud backup, for instance) is explicitly allowed. Good news if you want to keep primary data on-premise for cost and control while still meeting compliance.
• Security audit requirement: Getting production-ready under ABDM involves a CERT-In empanelled security audit. A server environment with proper RAID, access controls, and patch management goes into that audit in far better shape than one that's been running unmanaged for years.
• NABH's IMS (Information Management System) standard separately assesses medical records confidentiality, integrity, and security — which again comes back to access-controlled, redundant, properly backed-up server infrastructure, not just software policy on paper.

None of this changes what a PACS or HIS server fundamentally needs — it just raises the bar on redundancy and access control being done properly, not skipped to save cost.

New vs Refurbished — What Actually Makes Sense for Hospital IT

A lot of hospital administrators assume "medical-grade" means "has to be brand new" — that's not accurate. A PACS or HIS server is standard enterprise server hardware (Dell PowerEdge, HPE ProLiant/Apollo) running standard software; it carries no medical-device certification of its own — that sits with the imaging modality and the HIS software, not the server box.

What actually matters is that the hardware is properly tested and warranty-backed, which is exactly where refurbished enterprise servers earn their place in a hospital's budget:

• A tested, warrantied refurbished Dell PowerEdge R740 or HPE ProLiant DL380 Gen10 costs a fraction of new, while running the exact same dual-Xeon, RAID-10, dual-PSU configuration a hospital needs.
• For the PACS archive tier specifically, HPE Apollo servers and SAN storage for centralized multi-site archives deliver the density hospitals need at a much lower cost per TB than buying new.
• The honest caveat: refurbished only works when the hardware has actually been tested and comes with real warranty and AMC support — a "used server, as-is" purchase with no testing is a genuine risk for a system this critical.

Why Buy Your Hospital or Diagnostic Center Server from Serverwale

This is the gap Serverwale exists to close — every unit goes through a 72-point QC and testing process before it ships, with warranty and GST invoicing, specifically so a hospital or diagnostic center isn't gambling on a critical server the way an unverified used-hardware seller would have them do. We help size the right PACS storage and HIS server configuration for your actual bed count and imaging volume instead of selling you whatever's in stock, and every server ships with up to 3-year warranty with optional AMC so a drive failure gets caught before it turns into an outage. If budget is the reason you're considering refurbished at all, our refurbished server price index gives an honest sense of what a properly-specced PACS or HIS server actually costs before you commit.

Frequently Asked Questions

Can one server run both PACS and HIS for a small clinic? For a very small setup (under 20–25 beds, low imaging volume), it's technically possible on a single well-specced server with separate storage volumes and enough RAM to isolate the workloads. Beyond that, separate servers are strongly recommended — the workloads compete for resources in ways that cause real slowdowns once volume grows.

How long should PACS data actually be retained? Indian regulatory guidance generally points to a minimum of 3 years for most imaging records, though hospitals commonly retain 5–7 years, and some specialties (oncology, pediatrics) longer. Design your storage tiering — fast RAID for recent studies, cheaper archive storage for older ones — around your facility and specialty's actual retention requirement.

Does ABDM require us to host servers in a specific location or with a specific vendor? No. ABDM doesn't mandate where your servers physically sit or which vendor supplies them — it mandates the data formats (FHIR R4, DICOM, HL7) and the security audit process. On-premise hospital servers, including refurbished enterprise hardware, are fully compatible with ABDM compliance as long as security and interoperability requirements are met.

What's the real difference between RAID-5 and RAID-10 for a hospital server? RAID-5 tolerates one drive failure but has a slow, high-risk rebuild window where a second failure can mean data loss — not acceptable for a live patient database. RAID-10 mirrors and stripes data, rebuilds much faster, and tolerates certain multi-drive failures. For HIS databases and active PACS storage, RAID-10 is worth the extra drive cost.

Is server AMC really necessary if the hardware is under warranty? Warranty typically covers part replacement but not guaranteed response time or on-site support SLAs. For a hospital, a drive failure that takes three days to resolve is a real operational risk — AMC is what turns "warranty exists" into "someone shows up and fixes it today."

Setting Up Hospital IT Infrastructure? Talk to Us First

Sizing a PACS or HIS server wrong is an expensive mistake to fix later. Serverwale supplies tested, warranty-backed refurbished Dell, HPE, and Lenovo servers built for exactly this kind of always-on, can't-go-down environment, with server AMC to back it up. Call +91-87968-22044, WhatsApp us, or use our contact page for a configuration recommendation.