Paper Notes
1.
benchmarks
1.1.
ssb
2.
bigdata
2.1.
mapreduce
2.2.
nephele
2.3.
dataflow model
2.4.
flink
2.5.
flink state management
3.
databases
3.1.
cloudnative
3.1.1.
aurora
3.1.2.
taurus
3.2.
columnstores vs rowstores
3.3.
kv
3.3.1.
rocksdb cidr17
3.3.2.
wisckey
3.4.
mmdb
3.4.1.
mmdb overview
3.5.
oltp
3.5.1.
through the looking glass
3.5.2.
staring into the abyss
3.6.
olap
3.6.1.
lakehouse
3.6.2.
delta lake
3.6.3.
vertica
3.6.4.
duckdb
3.7.
htap
3.7.1.
greenplum
3.8.
vector db
3.8.1.
hnsw
3.8.2.
ivf-hnsw
3.8.3.
diskann
3.8.4.
product quantization
3.9.
citus
3.10.
optimizer
3.11.
executor
3.11.1.
volcano
3.12.
concurrency control
3.12.1.
evaluation of in-memory mvcc
3.13.
cdc
3.13.1.
dblog
3.14.
rum conjecture
4.
datalayout
4.1.
cstore
4.2.
cstore compression
4.3.
dremel
4.4.
rcfile
4.5.
orc
4.6.
table placement methods
5.
data structures
5.1.
btree family
5.1.1.
bw-tree
5.2.
hash table
5.2.1.
linear hashing
5.3.
trie family
5.3.1.
art
5.3.2.
hot
5.4.
bitmaps
5.4.1.
roaring bitmaps
5.5.
skip list
5.6.
bloom filter
6.
distributed system
6.1.
consensus
6.1.1.
flp
6.1.2.
paxos made simple
6.1.3.
paxos made live
6.1.4.
viewstamped replication
6.1.5.
zab
6.1.6.
paxos vs. vr vs. zab
6.1.7.
raft
6.1.8.
paxos vs raft
6.2.
scheduler
6.2.1.
borg
6.3.
primary backup
6.4.
chain replication
6.5.
bolosky
6.6.
holy grail
6.7.
chandy lamport
6.8.
asynchronous barrier snapshotting
6.9.
zookeeper
7.
filesystem
7.1.
gfs
7.2.
polarfs
8.
llm
9.
storage
9.1.
kv store
9.1.1.
dynamo
9.2.
kudu
9.3.
bluestore
Light (default)
Rust
Coal
Navy
Ayu
论文阅读笔记
filesystem
The Google File System
PolarFS: An Ultralow Latency and Failure Resilient Distributed File System for Shared Storage Cloud Database
Further readings
To FUSE or Not to FUSE: Performance of User-Space File Systems
, FAST ’17