Now syncing again. I copied bitcoin data to the slower VPS and started sqlchain on both systems processing blocks. Here is a snapshot of the first minutes on each.
First the cheaper ($5/mo.) VPS with 300GB RAID-10 storage:
4vCPU-4GB-300GB RAID-10 (MYI+MYD) (Hudson Valley Host)
sqlchain - daemon.log:
23-01-2017 18:39:17 Using rpc mode. Monitoring blocks / mempool.
Block 449396 [ Q:8 1434 txs - 22-01-2017 - 936ms 211.7s 7 tx/s]
Block 449397 [ Q:8 895 txs - 22-01-2017 - 722ms 67.3s 13 tx/s]
Block 449398 [ Q:8 343 txs - 22-01-2017 - 209ms 212.4s 2 tx/s]
Block 449399 [ Q:8 181 txs - 22-01-2017 - 94ms 31.9s 6 tx/s]
Block 449400 [ Q:8 359 txs - 22-01-2017 - 176ms 41.4s 9 tx/s]
Block 449401 [ Q:8 1094 txs - 22-01-2017 - 229ms 76.7s 14 tx/s]
Block 449402 [ Q:8 620 txs - 22-01-2017 - 141ms 200.2s 3 tx/s]
Block 449403 [ Q:8 2321 txs - 22-01-2017 - 480ms 164.5s 14 tx/s]
MySQL - show status (2048 MB key buffer)
| Key_blocks_not_flushed | 0 |
| Key_blocks_unused | 1690322 |
| Key_blocks_used | 24414 |
| Key_read_requests | 120304 |
| Key_reads | 24294 |
| Key_write_requests | 10368 |
| Key_writes | 10359 |
Now the SSD based VPS (~$27/mo, 45GB SSD + 75 GB block storage):
2vCPU-2GB-45GB SSD (MYI) + 75GB block storage (MYD) (Vultr)
sqlchain - daemon.log:
23-01-2017 11:40:24 Using rpc mode. Monitoring blocks / mempool.
Block 449396 [ Q:8 1434 txs - 21-01-2017 - 301ms 9.5s 150 tx/s]
Block 449397 [ Q:7 895 txs - 21-01-2017 - 228ms 3.1s 287 tx/s]
Block 449398 [ Q:8 343 txs - 21-01-2017 - 90ms 16.1s 21 tx/s]
Block 449399 [ Q:7 181 txs - 21-01-2017 - 66ms 1.9s 93 tx/s]
Block 449400 [ Q:6 359 txs - 21-01-2017 - 76ms 1.2s 305 tx/s]
Block 449401 [ Q:7 1094 txs - 21-01-2017 - 161ms 3.4s 319 tx/s]
Block 449402 [ Q:8 620 txs - 22-01-2017 - 92ms 5.3s 117 tx/s]
Block 449403 [ Q:8 2321 txs - 22-01-2017 - 395ms 6.6s 354 tx/s]
Block 449404 [ Q:8 1702 txs - 22-01-2017 - 2034ms 35.6s 48 tx/s]
Block 449405 [ Q:7 99 txs - 22-01-2017 - 49ms 0.4s 259 tx/s]
Block 449406 [ Q:8 1286 txs - 22-01-2017 - 483ms 12.5s 103 tx/s]
MySQL - show status (1200 MB key buffer)
| Key_blocks_not_flushed | 8626 |
| Key_blocks_unused | 499465 |
| Key_blocks_used | 507801 |
| Key_read_requests | 3064005 |
| Key_reads | 503180 |
| Key_write_requests | 325917 |
| Key_writes | 249240 |
Presumably less CPU power and RAM but ~ 20x block rate, though CPU power is hard to compare equivalence between VPS. On the hard disk system htop
shows CPU iowait bars about 5x higher than active CPU bars. The slower VPS does eventually start to get better as it's key buffer gets filled - it does have more memory and is configured to use it.
The SSD-VPS caught up to block 449,649 in 52 minutes by which time the HD-VPS was at block 449,413. Here's what the SSD-VPS looks like after catching up. Why so fast? It monitors the mempool and adds txs as they come in. So when a new block arrives it simply has to update the relevant txs with the new block number - very quick.
Block 449652 [ Q:0 1418 txs - 23-01-2017 - 289ms 0.5s 2582 tx/s]
Block 449653 [ Q:0 1521 txs - 23-01-2017 - 282ms 0.7s 2321 tx/s]
Block 449654 [ Q:0 1647 txs - 23-01-2017 - 372ms 3.6s 452 tx/s]
Block 449655 [ Q:0 2191 txs - 23-01-2017 - 373ms 1.1s 2001 tx/s]
Block 449656 [ Q:0 2382 txs - 23-01-2017 - 380ms 1.0s 2486 tx/s]
Block 449657 [ Q:0 2471 txs - 23-01-2017 - 416ms 1.0s 2419 tx/s]