@haohaojade: #phỉthuý #jade #ngọc #ngọcphỉthuý #phithuy

haohaojade
haohaojade
Open In TikTok:
Region: VN
Monday 24 August 2026 10:50:55 GMT
564
24
0
4

Music

Download

Comments

There are no more comments for this video.
To see more videos from user @haohaojade, please go to the Tikwm homepage.

Other Videos

Partitioning does not scale your database. That's the whole confusion. ⚡ Same 12 rows. Three arrangements. Count the dots and the difference becomes obvious 👇 🟢 1. PARTITIONING — manage big tables Split one table into chunks inside ONE database. 12 rows · 1 server · 1 disk 💥 Breaks when: the partition key doesn't match your query pattern. Then every query scans every partition and you gained nothing but complexity. 🔵 2. SHARDING — scale writes + storage Split data across SEPARATE machines. A hash decides which server owns which row. 4 rows each · 3 servers · 12 stored 💥 Breaks when: you need a join across shards, or you have to rebalance. 🟣 3. REPLICATION — survive failures Copy EVERY row to EVERY node. Nothing is split. 12 rows each · 3 servers · 36 stored 💥 Breaks when: you read right after a write. The replica is a few ms behind. 👀 The one nobody says: partitioning and sharding both store 12 rows. Replication stores 36. Partitioning uses one machine. The other two use three. Those two numbers separate all three completely — and they're why partitioning is the one people get wrong. It's still one CPU, one disk, one machine's write throughput. Splitting a table into partitions gives you query pruning and easier maintenance. Both real. Neither is scale. If you're partitioning because
Partitioning does not scale your database. That's the whole confusion. ⚡ Same 12 rows. Three arrangements. Count the dots and the difference becomes obvious 👇 🟢 1. PARTITIONING — manage big tables Split one table into chunks inside ONE database. 12 rows · 1 server · 1 disk 💥 Breaks when: the partition key doesn't match your query pattern. Then every query scans every partition and you gained nothing but complexity. 🔵 2. SHARDING — scale writes + storage Split data across SEPARATE machines. A hash decides which server owns which row. 4 rows each · 3 servers · 12 stored 💥 Breaks when: you need a join across shards, or you have to rebalance. 🟣 3. REPLICATION — survive failures Copy EVERY row to EVERY node. Nothing is split. 12 rows each · 3 servers · 36 stored 💥 Breaks when: you read right after a write. The replica is a few ms behind. 👀 The one nobody says: partitioning and sharding both store 12 rows. Replication stores 36. Partitioning uses one machine. The other two use three. Those two numbers separate all three completely — and they're why partitioning is the one people get wrong. It's still one CPU, one disk, one machine's write throughput. Splitting a table into partitions gives you query pruning and easier maintenance. Both real. Neither is scale. If you're partitioning because "the table got big and things are slow," check whether you actually needed a shard. 🎯 The rule: partition to manage. Shard to scale. Replicate to survive. And in production you run all three at once — sharded, each shard partitioned, each shard replicated. 📸 Screenshot it before your next system design round. Follow @hackproduct — we turn scary engineering concepts into things you can ship. ⚡ . . #systemdesign #databases #sharding #replication #postgres

About