Uploaded November 2025 | Updated September 2026, 2 weeks ago
Command Duration Limits (CDL) is a QoS protocol for SCSI and ATA HDDs that provides the host with a model of traffic classes and command execution policies that enable a drive to optimize execution of consumed commands. The standard has a two dimensional model. CDL defines Duration Limits by which the host can set general priority for classes of commands and it defines Duration Limit Policies to further guide the device on the desired behavior of commands as they approach the limits of the Duration Limits. CDL has been adopted as SCSI and ATA standards, and implemented by drives, the Linux kernel, and the SAS hardware ecosystem. It is also being considered for adoption for NVMe HDDs. In this talk, we will present the CDL behavioral model, and an overview of the standard as implemented in SCSI and ATA, as well as some thoughts on how this model would be adapted to NVMe. We will also present real world results of CDL’s implementation in the field, showing examples of its benefits on command tail latencies, and also the performance trade-offs involved in maximizing IOPs per TB in today’s ever higher capacity HDDs.
Provide the audience with a clear description of the CDL behavior model. Provide the audience with the performance improvements demonstrated by the implementation of CDL in the field Provide the audience with understanding of the state of support for CDL in the SW and HW ecosystem.
Presented by David Landsman, Western Digital and Damien Le Moal, Western Digital
Command Duration Limits (CDL) is a QoS protocol for SCSI and ATA HDDs that provides the host with a model of traffic classes and command execution policies that enable a drive to optimize execution of consumed commands. The standard has a two dimensional model. CDL defines Duration Limits by which the host can set general priority for classes of commands and it defines Duration Limit Policies to further guide the device on the desired behavior of commands as they approach the limits of the Duration Limits. CDL has been adopted as SCSI and ATA standards, and implemented by drives, the Linux kernel, and the SAS hardware ecosystem. It is also being considered for adoption for NVMe HDDs. In this talk, we will present the CDL behavioral model, and an overview of the standard as implemented in SCSI and ATA, as well as some thoughts on how this model would be adapted to NVMe. We will also present real world results of CDL’s implementation in the field, showing examples of its benefits on command tail latencies, and also the performance trade-offs involved in maximizing IOPs per TB in today’s ever higher capacity HDDs.
Provide the audience with a clear description of the CDL behavior model. Provide the audience with the performance improvements demonstrated by the implementation of CDL in the field Provide the audience with understanding of the state of support for CDL in the SW and HW ecosystem.
Presented by David Landsman, Western Digital and Damien Le Moal, Western Digital









![DNA MGC+ A Codec for Reliable and Efficient DNA Data Storage
Efficient and reliable data retrieval remains a major challenge in DNA data storage due to the inherent noisiness of the underlying biochemical processes, which lead to both base-level errors and sequence-level dropouts. Here we introduce DNA-MGC+, a novel DNA storage codec designed to enable reliable and efficient data retrieval in the presence of insertion, deletion, and substitution (IDS) errors as well as dropouts. DNA-MGC+ combines an inner coding layer based on the Marker Guess & Check Plus (MGC+) code [1] for correcting IDS errors with an outer Reed-Solomon code that recovers from sequence dropouts and corrects residual inner decoding errors. Our results show that DNA-MGC+ consistently outperforms other codecs across diverse operating conditions. In particular, we observe gains in sequencing depth requirements and decoding time under both Illumina and Nanopore sequencing. We evaluated the performance of DNA-MGC+ in comparison with representative codecs through an in vitro experiment in which sequences encoded using multiple codec configurations were combined in a single oligonucleotide pool. Specifically, a 24-KB compressed file was encoded into oligonucleotides of length 170 bases using two configurations of DNA-MGC+ with different redundancy allocations, design A (1.03 bits/nt) and design B (0.71 bits/nt), as well as two existing codecs: DNA-Aeon [2] (1 bit/nt) and HEDGES [3] (0.61 bits/nt). The oligonucleotide pool was ordered from GenScript (electrochemical synthesis) and sequenced using both Illumina and Oxford Nanopore platforms, with multiple basecalling algorithms evaluated for the Nanopore data. Across all sequencing and basecalling setups, the stored file was recovered with an exact match, albeit with quantitatively different performance outcomes. The results shown in the attached figure indicate that DNA-MGC+ consistently outperforms both DNA-Aeon and HEDGES in terms of the minimum sequencing depth required for reliable decoding, achieving depths below 3x for both Illumina and Nanopore sequencing.
Presented by
Serge Kas Hanna, CNRS
This is a presentation from the 2026 Storage and Computing with DNA Conference.
· Learn More about the SNIA DNA Data Storage Alliance: https://www.snia.org/groups/snia-dna-technology-affiliate
· SNIA Educational Library: https://snia.org/library
· X: https://twitter.com/SNIA
· LinkedIn: https://linkedin.com/company/snia/ DNA MGC+ A Codec for Reliable and Efficient DNA Data Storage](https://i.ytimg.com/vi/gqRbmqRlTMM/mqdefault.jpg)
