Lets Talk About Storage & Recovery Methods for Non-Volatile - PowerPoint PPT Presentation

Let’s Talk About Storage & Recovery Methods for Non-Volatile Memory OLTP Database Systems Andy Pavlo + Joy Arulraj Carnegie Mellon University

Winter Winter 2013: 2013: Fir irst B t Blood lood • Initial evaluation of existing DBMSs on Intel NVM SDV • Results published ADMS@VLDB’14 2

MyS MySQL L vs. vs. H H-Sto Store 90% Reads / 10% Writes 50% Reads / 50% Writes 3

Summer Summer 2014 2014: : Fir irst t Blood lood, P , Par art II t II • Evaluate storage and recovery methods for NVM. • Preparing SIGMOD’14 submission. 4

DBMS DBMS Tes estbe tbed • Custom lightweight DBMS. – Uses NUMA & PMFS interfaces. – No volatile DRAM. • Partition-based locking CC. • Pluggable architecture: – Supports different storage engines. 5

Engine ngine #1 #1 – In In-pla place ce U Upda pdate tes • Apply change to tuples directly. – VoltDB with ARIES. – Table storage + write-ahead log. – STX B+Tree 6

Engine ngine #2 #2 – Copy opy-on on-Write U ite Upda pdate tes • Make new copy before updating: – Shadow paging using LMDB Persistent B+Tree. – No logging. – Background garbage collection. 7

Engine ngine #3 #3 – Log og-bas based U ed Updat ates es • Changes only written to log. – Based on LevelDB’s LSM. – No table storage. – Background level compaction. 8

Stor Storage E Engin ngines Tabl able St Stor orag age Loggi gging Examp Ex mple In-Place VoltDB Yes Yes Copy-on-Write Yes No LMDB Log-based No Yes LevelDB 9

NVM O Optimiz ptimized E Engine ngines • Refactored engines to be “pointer-oriented”. • Extended Intel’s libpmem allocation library. – Added arena-based allocation. – Significantly improved throughput. 10

Expe xperime imenta tal E l Evalua aluatio tion • Yahoo! Cloud Serving Benchmark: – 2 million records (~2GB) – Two workload mixtures – Two skew settings – 1 million transactions 11

Expe xperime imenta tal E l Evalua aluatio tion • NVM Latency Configuration: – 2x DRAM (~200ns) – 8x results not shown. • 8 partitions on 8 cores. 12

Throughput oughput 4x 4x 90% Reads / 10% Writes 50% Reads / 50% Writes 13

NVM R Read eads/Wr Write ites Stores Loads 90% Reads / 10% Writes 50% Reads / 50% Writes 14

Reco ecovery T Time ime Zero Recovery 15

Dis iscu cussio ion • NVM engines outperforms “traditional” engines: – Higher throughput – Reduced wear on device. • In-place performs best overall. 16

N-STORE STORE nstore.cs.cmu.edu

Fall all 2014 2014: : N-Sto Store • First DBMS for NVM-only operating environment. • OLTP/OLAP hybrid – Column-store that supports fast in- place updates. • Indexing + Many-Core 18

Justin Joy Rajesh Subramanya DeBrabant Arulraj Sankaran Dulloor Andy Mike Col. Stan Jeff Pavlo Stonebraker Zdonik Parkhurst

END @ ANDY _ PAVLO