\u00a0<\/p>\r\n
\u00a0Program<\/h2>\r\n
![\"\"](\"https:\/\/www.systor.org\/2025\/wp-content\/uploads\/sites\/9\/2025\/07\/ISW_sched-300x220.png\")
<\/p>\r\n
\u00a0<\/p>\r\n
Keynote<\/h2>\r\n
Scaling LSM-Trees for the AI Age and Beyond<\/span><\/strong><\/p>\r\n Niv Dayan (University of Toronto)<\/strong><\/em><\/p>\r\n Abstract<\/strong>: This talk revisits core LSM-tree design principles and presents recent advances that upend longstanding trade-offs between read and write efficiency. We discuss the use of range filters as a replacement for traditional Bloom filters and introduce techniques for optimizing filter space allocation across levels to reduce false positives and compaction costs simultaneously. These techniques unlock new regimes of performance, enabling LSM-based systems to scale more gracefully under modern demands.<\/p>\r\n Bio<\/strong>: \u00a0<\/p>\r\n Chair: Joel Nider <\/strong>(Unifabrix)<\/p>\r\n Chair: Amit Golander <\/strong>(Tel-Aviv University, PLiOPS)<\/p>\r\n Chair: Doron Chen <\/strong>(IBM Research)<\/p>\r\n \u00a0<\/p>\r\n The IBM Haifa facility is located at the northwest edge of the University of Haifa campus.<\/p>\r\n Parking is available at IBM.<\/p>\r\n![\"\"](\"https:\/\/www.systor.org\/2025\/wp-content\/uploads\/sites\/9\/2025\/06\/NivDayan-300x200.jpeg\")
<\/p>\r\n
Log-Structured Merge-trees (LSM-trees) underpin large-scale storage systems such as Bigtable, DynamoDB, and RocksDB, and are foundational to ML and AI infrastructure including feature stores, recommendation engines, and vector databases. While their write-optimized design is well-suited to write-intensive workloads, scaling both read and write performance simultaneously remains a central challenge.<\/p>\r\n
Niv Dayan is an Assistant Professor at the University of Toronto (U of T). His research focuses on designing and analyzing data structures for database and storage systems. Before joining U of T, he was a Research Scientist at Pliops and a Technical Advisor to Speedb. He received his Ph.D. from the IT University of Copenhagen and was a postdoctoral researcher at Harvard.<\/p>\r\nTalks<\/h2>\r\n
Highlight Papers \u2013 Session A<\/h3>\r\n
\r\n
<\/strong>Presented by: Gal Dalal <\/strong>(Nvidia Research)
Congestion control<\/a>, GBRL\u00a0<\/a>
Slides<\/a><\/li>\r\n
<\/strong>Presented by: Kfir Y. Levy <\/strong>(Technion ECE)\u00a0
<\/strong>One of the papers covered<\/a>
Slides<\/a><\/li>\r\n
<\/strong>Presented by: Vita Bortnikov <\/strong>(IBM Research)\u00a0
<\/strong>llm-d<\/a>
Slides<\/a><\/li>\r\n<\/ul>\r\nSystems Optimized for AI \u2013 Session B<\/h3>\r\n
\r\n
Hezi Rahamim<\/strong>\u00a0(Google)<\/li>\r\n
<\/strong>Eshcar Hillel <\/strong>(PLiOPS)
Slides<\/a><\/li>\r\n
<\/strong><\/strong>Danny Volkind <\/strong>(Unifabrix)
Slides<\/a><\/li>\r\n
Ilan Tayari <\/strong>(NextSilicon)<\/li>\r\n<\/ul>\r\nAgentic AI \u2013 Session C<\/h3>\r\n
\r\n
Gil Vernik<\/strong>\u00a0<\/strong>(IBM Research)
Slides<\/a><\/li>\r\n
Sarit Kraus <\/strong>(Bar-Ilan University)
Slides<\/a><\/li>\r\n
<\/strong>Sagi Grimberg <\/strong>(VAST Data)
Slides<\/a><\/li>\r\n<\/ul>\r\nVenue<\/h2>\r\n