Thousands of Duplicate Images Are Clogging Dhaka's Government Databases — and the Numbers Tell a Damning Story

At least 34 percent of image files stored across Dhaka's municipal digital infrastructure are duplicates — identical or near-identical copies that consume server space, inflate IT costs, and slow the retrieval systems that civil servants rely on daily, according to internal assessments shared with The Daily Dhaka by municipal technology officials who declined to be named in full.

The problem has sharpened this year as Dhaka North City Corporation (DNCC) and Dhaka South City Corporation (DSCC) both push forward with digitisation drives meant to modernise land records, voter registration, and public health databases. What those drives have exposed is an unglamorous but expensive legacy problem: years of scanning paper documents without deduplication protocols have left government servers groaning under the weight of redundant files.

The Scale of the Problem in Numbers

Bangladesh Computer Council's e-Government cell estimated in a 2025 review that government agencies collectively spend upward of Tk 180 crore annually on data storage procurement — a figure that officials say could fall by as much as 20 percent if basic image deduplication were enforced across ministries. The DSCC alone, headquartered at Nagar Bhaban on Fulbaria Road, holds an image archive now exceeding 14 terabytes, a volume that has tripled since 2021.

Part of the explosion traces back to the National ID digitisation programme managed through the Election Commission's offices on Agargaon. When physical voter cards and supporting documents were scanned in bulk between 2020 and 2023, operators frequently rescanned pages after mechanical errors, then stored both versions. No automated flag removed the redundant copy. A 2024 spot-check cited in the Bangladesh Institute of ICT in Development's annual report found duplication rates in scanned civic documents running between 28 and 41 percent depending on the district office.

The problem is not purely a government one. The Dhaka metropolitan area's private sector — particularly garment sourcing offices concentrated in Gulshan-1 and Banani — has faced parallel issues as teams managing fabric samples and compliance photography discovered their shared drives were carrying two, three, or even four versions of the same product image. One mid-sized sourcing firm operating out of Gulshan Avenue reported internally that a single clean-up exercise in March 2025 recovered 1.8 terabytes of storage that had been paid for at roughly Tk 4,200 per terabyte per year on a cloud contract.

Why Deduplication Has Been Slow to Arrive

Procurement rules are part of the answer. Under current government IT tender frameworks, agencies specify storage capacity in raw terabyte terms rather than effective usable capacity after deduplication. That creates a perverse incentive: vendors sell more storage, and the agencies buying it have no contractual reason to clean what they already hold.

The Access to Information programme, operating under the Cabinet Division's a2i unit based in the Bangladesh Secretariat compound off Abdul Gani Road, has been piloting a document management system called DMS-BD since late 2024. Early results from two upazila offices in Mirpur showed a 22 percent reduction in indexed file size within six months of switching on hash-based deduplication — a process that identifies and removes bit-for-bit identical copies before they are committed to permanent storage.

Wider rollout has stalled, however, pending a budget line in the next Annual Development Programme cycle. Technology officers at both city corporations say they are watching the a2i pilot closely but cannot proceed independently without central guidance on data governance standards.

For ordinary Dhaka residents, the practical downstream effect is recognisable even if the technical cause is not: slow loading times on the DSCC property tax portal, delays in retrieving scanned birth certificates at ward offices in Lalbagh and Hazaribagh, and recurring errors when the same document image is updated in one database but not in its duplicate stored elsewhere.

Civic technology advocates say the fix is neither complicated nor prohibitively expensive. Open-source deduplication tools are available free of charge, and the Bangladesh Computer Council has the technical staff to deploy them. What the city needs, they argue, is a firm ministerial deadline — something that has so far not appeared in any budget speech or digital policy circular — and a clear data-ownership framework before the next major scanning drive begins.