Data EngineeringNOV 6, 2026
Small Files and Too Many Parts: Why Batching Your Inserts Matters
Tiny writes are cheap to send and expensive to keep. We wrote the same 2 million rows as 20,000 files and as 2: 2.9× the bytes, 54× the read time.…
A new essay every morning
Byte2MB is Priyansh Nigam’s daily newsletter on storage, databases, observability and research, with calculators for the size questions that come up in design reviews.
Decimal on the left, binary on the right. They differ by 7.4% at a gigabyte.
SI decimal and IEC binary
No upload, no sign-up
Shrink a photo or signature to 20, 50 or 100 KB.
Target 20, 50, 100 or 200 KBPhoto and signature, under 20 KB.
Photo or signature, under 20 KBShrink a PDF under 200 KB, no upload.
Any PDF, under 200 KBBytes into megabytes, decimal and binary.
1,500,000 B → 1.5 MBDivide by 1,000 or by 1,024.
1,536 KB → 1.536 MBThe 1,024 trap, labeled.
2,560 MB → 2.56 GBData EngineeringNOV 6, 2026
Tiny writes are cheap to send and expensive to keep. We wrote the same 2 million rows as 20,000 files and as 2: 2.9× the bytes, 54× the read time.…
Measured
Research notes will land here.