On this page
Why the internal network matters
An internal test answers one question: if one laptop is phished or one vendor account is stolen, how far can an attacker get? In a hospital the answer decides whether an incident is a single infected PC or ransomware across clinical systems.
The work combined a credentialed vulnerability assessment of the responsive hosts, a grey-box penetration test from inside the network to confirm what was exploitable, and a database assessment of the main Oracle server, following NIST SP 800-115 with non-destructive checks in approved windows.
What we found
- Windows remote code execution flaws: BlueKeep (Remote Desktop), SIGRed (Windows DNS Server), QueueJumper (Message Queuing) and EternalBlue (SMB), each able to give an attacker control of a server without a login.
- PrintNightmare and weak service permissions, both routes to higher privileges.
- Legacy protocols and authentication: SSLv2 and SSLv3, rlogin and Telnet sending credentials in clear text, and LM and NTLMv1 still enabled, with exposure to NTLM relay.
- Unsupported and unpatched software across the estate, including outdated infrastructure components.
- A core database behind on Oracle Critical Patch Updates.
Together these give an attacker the classic path from one foothold to domain-wide control.
None of this needed a zero-day. This is the most common pattern we see in internal tests across the Gulf: the risk is not unknown vulnerabilities, it is known ones that were never closed.
The database
For a hospital, the core database is where patient, billing and clinical records meet. It should be the best-patched, most restricted system on the network, reachable only from the application servers that need it. Here it had fallen behind on vendor updates that fix remotely exploitable flaws.
What to fix first
- Patch or isolate hosts exposed to BlueKeep, SIGRed and EternalBlue, and disable SMBv1 wherever it remains
- Retire or replace unsupported, unpatched software
- Apply the missing database updates and restrict which hosts can reach the database
- Disable SSLv2, SSLv3, rlogin and Telnet
- Turn off LM and NTLMv1 and protect against NTLM relay
Every item is closed only after a retest proves the fix.
Lessons for Gulf hospitals and enterprises
- Inventory first. You cannot patch what you have not listed, including vendor-managed medical systems.
- Segment what cannot be patched. Imaging and device servers that must stay on old Windows belong in their own zone.
- Kill legacy authentication. LM, NTLMv1 and clear-text protocols turn one stolen password into many.
- Treat the database as the crown jewel. Regular patches, minimal network reach, monitored access.
See Sumrite’s network penetration testing and red team operations, or read this case study in Arabic.
Frequently asked questions
Why do hospitals still have BlueKeep and EternalBlue exposures?
Medical devices, imaging systems and vendor-managed servers often run older Windows versions that cannot be patched without the vendor's approval. They stay on the network, unsegmented, long after patches exist. The fix is an inventory, vendor pressure, and segmentation for anything that cannot be patched.
What is the difference between an internal vulnerability assessment and a penetration test?
The vulnerability assessment is a credentialed scan that lists known weaknesses on each host. The penetration test confirms which ones can actually be exploited from a foothold inside the network and what they lead to. This engagement ran both.
How often should a hospital patch its Oracle database?
Oracle releases Critical Patch Updates every quarter. A database that falls behind carries vulnerabilities that can be exploited remotely, some without a login. Apply each update after testing, and restrict which hosts can reach the database at all.
Is an internal test disruptive?
It does not have to be. Sumrite uses non-destructive, grey-box methods in agreed windows and confirms exploitability without crashing services. Destructive checks are only run with written approval.
