Arif Merchant
Arif Merchant is a Research Scientist with the Storage Analytics group at Google, where he studies interactions between components of the storage stack. Prior to this, he was with HP Labs, where he worked on storage QoS, distributed storage systems, and stochastic models of storage. He holds the B.Tech. degree from IIT Bombay and the Ph.D. in Computer Science from Stanford University. He is an ACM Distinguished Scientist.
Research Areas
Authored Publications
Sort By
TCO-driven Storage Provisioning for Exascale Data Centers
Timothy Kim
Prashant Nema
Jai Menon
Gregory R. Ganger
Rashmi Vinayak
2026
Preview abstract
Recent changes in data temperatures and storage device characteristics, both mechanical disk drives (HDDs) and solidstate drives (SSDs), expand the set of deployment options for exascale storage. Until recently, exascale storage systems followed a pattern of placing most data on HDDs with smaller amounts of SSD storage used for caching and performance-critical workloads. Exascale storage provisioning and dataset placement trade-offs have now changed.
This paper describes a total cost of ownership (TCO) model that captures primary aspects of modern deployments and uses it to explore the new trade-off space. Using capacity and performance telemetry information for 43 production datasets+workloads at two large hyperscalers, we show significant changes from prior analyses of workloads and storage placement decisions across a multitude of storage device types. We also introduce a storage cluster TCO optimizer that identifies the lowest-TCO grouping and assignment of datasets to device types, exposing a number of insights that can help guide future deployments. For example, our analysis shows that the highest-density SSDs are particularly favorable for clusters with heavy AI/ML workloads but are only cost-effective at exascale when combined with high-density HDDs. Finally, we use our framework to evaluate how storage provisioning and overall TCO change as a function of key parameters like device write amplification, cluster power bounds, and the maximum number of device types allowed.
View details
Okapi: Decoupling data striping from grouping in cluster storage systems
Sanjith Athlur
Timothy Kim
Francisco Maturana
Xavier Ramos
Rashmi KV
Gregory R. Ganger
2025
Preview abstract
The Okapi cluster file system decouples how data is spread across disks (data striping) for IO efficiency from how data is erasure coded together (redundancy grouping) for durability. Existing systems couple these two mechanisms’ configurations, inducing significant inefficiencies. Decoupling allows grouping to be configured based on reliability and space efficiency goals, while simultaneously allowing striping to be configured based on performance goals. Decoupling also allows redundancy scheme changes from one EC scheme to another (e.g., to react to data temperature or disk failure rate changes) to occur without having to re-write data. Evaluation of an Okapi prototype shows that decoupling can be accomplished with <1% increase in metadata size and file manager memory, and minimal file creation and degraded read resource increase. Experiments demonstrate that decoupling can improve read throughput by 80% and reduce seeks per second by up to 70%, without yielding any data reliability, and reduce the overhead of redundancy transitions by up to 70%.
View details
Morph: Efficient File-Lifetime Redundancy Management for Cluster File Systems
Timothy Kim
Sanjith Athlur
Francisco Maturana
Alessandra Vandevoorde
Rashmi KV
Gregory R. Ganger
SOSP (2024)
Preview abstract
Many data services tune and change redundancy configurations of files over their lifetimes to address changes in data temperature and latency requirements. Unfortunately, changing redundancy configs (transcode) is IO-intensive. The Morph cluster file system introduces new transcode-efficient redundancy schemes to minimize overheads as files progress through lifetime phases. For newly ingested data, commonly stored via 3-way replication, Morph introduces a hybrid redundancy scheme that combines a replica with an erasure-coded (EC) stripe, reducing both ingest IO and capacity overheads while enabling free transcode to EC by deleting replicas. For subsequent transcodes to wider, more space-efficient EC configs, Morph exploits Convertible Codes, which minimize data read for EC transcode, and introduces new block placement policies to maximize their effectiveness.
Analysis of data ingest and transcode activity in Google storage clusters shows the current massive IO load and the potential savings from Morph's approach---transcode IO can be reduced by over 95%, and total ingest+transcode IO can be reduced by 50--60% while also reducing capacity overheads for newly ingested data by 20%. Experiments evaluating a Morph implementation in HDFS show that these benefits can be realized in a real system without hidden increases in complexity, tail latency, or degraded-mode latency.
View details
Practical Design Considerations for Wide Locally Recoverable Codes (LRCs)
Shashwat Silas
Dave Clausen
File and Storage Technologies (FAST), USENIX (2023)
Preview abstract
Most of the data in large-scale storage clusters is erasure coded. At exascale, optimizing erasure codes for low storage overhead, efficient reconstruction, and easy deployment is of critical importance. Locally recoverable codes (LRCs) have deservedly gained central importance in this field, because they can balance many of these requirements. In our work we study wide LRCs; LRCs with large number of blocks per stripe and low storage overhead. These codes are a natural next step for practitioners to unlock higher storage savings, but they come with their own challenges. Of particular interest is their reliability, since wider stripes are prone to more simultaneous failures.
We conduct a practically-minded analysis of several popular and novel LRCs. We find that wide LRC reliability is a subtle phenomenon that is sensitive to several design choices, some of which are overlooked by theoreticians, and others by practitioners. Based on these insights, we construct novel LRCs called Uniform Cauchy LRCs, which show excellent performance in simulations, and a 33% improvement in reliability on unavailability events observed by a wide LRC deployed in a Google storage cluster. We also show that these codes are easy to deploy in a manner that improves their robustness to common maintenance events. Along the way, we also give a remarkably simple and novel construction of distance optimal LRCs (other constructions are also known), which may be of interest to theory-minded readers.
View details
Tiger: disk-adaptive redundancy without placement restrictions
Francisco Maturana
Sanjith Athlur
Rashmi KV
Gregory R. Ganger
Tiger: disk-adaptive redundancy without placement restrictions (2022)
Preview abstract
Large-scale cluster storage systems use redundancy (via erasure coding) to ensure data durability. Disk-adaptive redundancy—dynamically tailoring the redundancy scheme to observed disk failure rates—promises significant space and cost savings. Existing disk-adaptive redundancy systems, however, pose undesirable constraints on data placement, partitioning disks into subclusters with homogeneous failure rates and forcing each erasure-coded stripe to be entirely placed on the disks within one subcluster. This design increases risk, by reducing intra-stripe diversity and being more susceptible to unanticipated changes in a make/model’s failure rate, and only works for very large storage clusters fully committed to
disk-adaptive redundancy.
Tiger is a new disk-adaptive redundancy system that efficiently avoids adoption-blocking placement constraints, while also providing higher space-savings and lower risk relative to prior designs. To do so, Tiger introduces the eclectic stripe, in which disks with different failure rates can be used to store a stripe that has redundancy tailored to the set of failure rates of those disks. With eclectic stripes, pre-existing placement policies can be used while still enjoying the space-savings and robustness benefits of disk-adaptive redundancy. This paper introduces eclectic striping and Tiger’s design, including a new mean time-to-data-loss (MTTDL) approximation technique and new approaches for ensuring safe per-stripe settings given that failure rates of different devices change over time. Evaluation with logs from real-world clusters show that Tiger provides better space-savings, less bursty IO for changing redundancy schemes, and better robustness (due to increased risk-diversity) than prior disk-adaptive redundancy designs.
View details
CacheSack: Admission Optimization for Datacenter Flash Caches
Homer Wolfmeister
Mustafa Uysal
Seth Bradley Pollen
Tzu-Wei Yang
2022 USENIX Annual Technical Conference (2022) (to appear)
Preview abstract
This paper describes the algorithm, implementation, and deployment experience of CacheSack, the admission algorithm for Google datacenter flash caches. CacheSack minimizes the dominant costs of Google’s datacenter flash caches: disk IO and flash footprint. CacheSack partitions cache traffic into disjoint categories, analyzes the observed cache benefit of each subset, and formulates a knapsack problem to assign the optimal admission policy to each subset. Prior to this work, Google datacenter flash cache admission policies were optimized manually, with most caches using the Lazy Adaptive Replacement Cache (LARC) algorithm. Production experiments showed that CacheSack significantly outperforms the prior static admission policies for a 6.5% improvement of the total operational cost, as well as significant improvements in disk reads and flash wearout.
View details
Preview abstract
Solid-state drives (SSDs) based on NAND flash are making deep inroads into data centers as well as the consumer market. In 2016, manufacturers shipped more than 130 million units totaling around 50 Exabytes of storage capacity. As the amount of data stored on solid state drives keeps increasing, it is important to understand the reliability characteristics of these devices. For a long time, our knowledge about flash reliability was derived from controlled experiments in lab environments under synthetic workloads, often using methods for accelerated testing. However, within the last two years, three large-scale field studies have been published that report on the failure behavior of flash devices in production environments subjected to real workloads and operating conditions. The goal of this paper is to provide an overview of what we have learned about flash reliability in production, and where appropriate contrasting it with prior studies performing controlled experiments.
View details
Preview abstract
The use of solid state drives based on NAND flash technology is continuously growing. As more data either lives on flash or is being cached on flash, data durability and availability critically depend on flash reliability.
This paper provides a detailed field study of flash reliability based on data collected over 6 years in a large-scale data center production environment. The data spans many millions of drive days, ten different drive models, different flash technologies (MLC and SLC) and feature sizes (ranging from 24nm to 50nm). The paper analyses this data in order to derive a better understanding of flash reliability in the field, including the most prevalent types of errors and hardware failures and their frequency, and how different factors impact flash reliability.
View details
Slicer: Auto-Sharding for Datacenter Applications
Atul Adya
Jon Howell
Jeremy Elson
Colin Meek
Vishesh Khemani
Stefan Fulger
Pan Gu
Lakshminath Bhuvanagiri
Jason Hunter
Roberto Peon
Alexander Shraer
Kfir Lev-Ari
OSDI 2016 (2016)
Preview abstract
Sharding is a fundamental building block of large-scale applications, but most have their own
custom, ad-hoc implementations. Our goal is to make sharding as easily reusable as a filesystem or
lock manager. Slicer is \Google's general purpose sharding service. It monitors signals such as
load hotspots and server health and dynamically shards work over a set of servers. Its goals are to
maintain high availability and reduce load imbalance while minimizing churn from moved work.
In this paper, we describe Slicer's design and implementation. Slicer has the consistency and global
optimization of a centralized sharder while approaching the high availability, scalability, and low
latency of systems that make local decisions.
It achieves this by separating concerns:
a reliable data plane forwards requests, and
a smart control plane makes load-balancing decisions off the critical path.
Slicer's small but powerful API has proven useful and easy to adopt
in dozens of \Google applications.
It is used to allocate resources for web service front-ends,
coalesce writes to increase storage bandwidth, and
increase the efficiency of a web cache.
It currently handles 2-6M~req/s of production traffic.
Production workloads using Slicer
exhibit a most-loaded task 30\%--180\% of the mean load,
even for highly skewed and time-varying loads.
View details
Poster Paper: Automatic Reconfiguration of Distributed Storage
Artyom Sharov
Alexander Shraer
Murray Stokely
The 12th International Conference on Autonomic Computing, IEEE (2015), pp. 133-134
Preview abstract
The configuration of a distributed storage system with multiple data replicas typically includes the set of servers and their roles in the replication protocol. The configuration can usually be changed manually, but in most cases, system administrators have to determine a good configuration by trial and error. We describe a new workload-driven optimization framework that dynamically determines the optimal configuration at run time. Applying the framework to a large-scale distributed storage system used internally in Google resulted in halving the operation
latency in 17% of the tested databases, and reducing it by more than 90% in some cases.
View details
