Skip to content

Commit d5116c4

Browse files
ekpghjswoodward
authored andcommitted
avere-vfxt: changes from periodic review
1 parent c7d1c30 commit d5116c4

15 files changed

+73
-83
lines changed

articles/avere-vfxt/avere-vfxt-additional-resources.md

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -4,7 +4,7 @@ description: Links to additional information about Avere vFXT for Azure
44
author: ekpgh
55
ms.service: avere-vfxt
66
ms.topic: conceptual
7-
ms.date: 01/09/2019
7+
ms.date: 12/19/2019
88
ms.author: rohogue
99
---
1010

@@ -28,4 +28,4 @@ Additional Avere cluster documentation can be found on the website at <https://a
2828

2929
## vFXT creation and management documentation
3030

31-
A full guide for using vfxt.py, the cloud cluster creation and management utility, is provided on GitHub: [Cloud cluster management with vfxt.py](https://github.com/Azure/AvereSDK/blob/master/docs/README.md).
31+
A full guide for using vfxt.py, a script-based cloud cluster creation and management utility, is provided on GitHub: [Cloud cluster management with vfxt.py](https://github.com/Azure/AvereSDK/blob/master/docs/README.md).

articles/avere-vfxt/avere-vfxt-configure-dns.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -4,7 +4,7 @@ description: Configuring a DNS server for round-robin load balancing with Avere
44
author: ekpgh
55
ms.service: avere-vfxt
66
ms.topic: conceptual
7-
ms.date: 10/31/2018
7+
ms.date: 12/19/2019
88
ms.author: rohogue
99
---
1010

articles/avere-vfxt/avere-vfxt-data-ingest.md

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -4,15 +4,15 @@ description: How to add data to a new storage volume for use with the Avere vFXT
44
author: ekpgh
55
ms.service: avere-vfxt
66
ms.topic: conceptual
7-
ms.date: 11/21/2019
7+
ms.date: 12/16/2019
88
ms.author: rohogue
99
---
1010

1111
# Moving data to the vFXT cluster - Parallel data ingest
1212

13-
After you've created a new vFXT cluster, your first task might be to move data onto its new storage volume. However, if your usual method of moving data is issuing a simple copy command from one client, you will likely see a slow copy performance. Single-threaded copying is not a good option for copying data to the Avere vFXT cluster's backend storage.
13+
After you've created a new vFXT cluster, your first task might be to move data onto a new storage volume in Azure. However, if your usual method of moving data is issuing a simple copy command from one client, you will likely see a slow copy performance. Single-threaded copying is not a good option for copying data to the Avere vFXT cluster's backend storage.
1414

15-
Because the Avere vFXT cluster is a scalable multi-client cache, the fastest and most efficient way to copy data to it is with multiple clients. This technique parallelizes ingestion of the files and objects.
15+
Because the Avere vFXT for Azure cluster is a scalable multi-client cache, the fastest and most efficient way to copy data to it is with multiple clients. This technique parallelizes ingestion of the files and objects.
1616

1717
![Diagram showing multi-client, multi-threaded data movement: At the top left, an icon for on-premises hardware storage has multiple arrows coming from it. The arrows point to four client machines. From each client machine three arrows point toward the Avere vFXT. From the Avere vFXT, multiple arrows point to Blob storage.](media/avere-vfxt-parallel-ingest.png)
1818

@@ -39,12 +39,12 @@ The data ingestor VM is part of a tutorial where the newly created VM mounts the
3939

4040
## Strategic planning
4141

42-
When building a strategy to copy data in parallel, you should understand the tradeoffs in file size, file count, and directory depth.
42+
When designing a strategy to copy data in parallel, you should understand the tradeoffs in file size, file count, and directory depth.
4343

4444
* When files are small, the metric of interest is files per second.
4545
* When files are large (10MiBi or greater), the metric of interest is bytes per second.
4646

47-
Each copy process has a throughput rate and a files-transferred rate, which can be measured by timing the length of the copy command and factoring the file size and file count. Explaining how to measure the rates is outside the scope of this document, but it is imperative to understand whether you’ll be dealing with small or large files.
47+
Each copy process has a throughput rate and a files-transferred rate, which can be measured by timing the length of the copy command and factoring the file size and file count. Explaining how to measure the rates is outside the scope of this document, but it is important to understand whether you’ll be dealing with small or large files.
4848

4949
## Manual copy example
5050

articles/avere-vfxt/avere-vfxt-demo-links.md

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -4,7 +4,7 @@ description: links to code demonstration tutorials for Avere vFXT
44
author: ekpgh
55
ms.service: avere-vfxt
66
ms.topic: conceptual
7-
ms.date: 10/31/2018
7+
ms.date: 12/19/2019
88
ms.author: rohogue
99
---
1010

@@ -14,13 +14,13 @@ Sample tutorials are provided on [GitHub](https://github.com/Azure/Avere). These
1414

1515
## Video rendering
1616

17-
* [Rendering using Azure Batch and Avere vFXT](https://github.com/Azure/Avere/blob/master/docs/maya_azure_batch_avere_vfxt_demo.md) - A 60-minute demo that demonstrates how to use Autodesk Maya with Azure Batch and an Avere vFXT cluster to generate an animated movie
17+
* [Rendering using Azure Batch and Avere vFXT](https://github.com/Azure/Avere/blob/master/docs/maya_azure_batch_avere_vfxt_demo.md) - A 60-minute project that demonstrates how to use Autodesk Maya with Azure Batch and an Avere vFXT cluster to generate an animated movie
1818

19-
* [Why use the Avere vFXT for rendering?](https://github.com/Azure/Avere/blob/master/docs/why_avere_for_rendering.md) - A demonstration that compares rendering times from NAS with and without an Avere vFXT cluster
19+
* [Why use the Avere vFXT for rendering?](https://github.com/Azure/Avere/blob/master/docs/why_avere_for_rendering.md) - A demonstration that compares rendering times from network-attached storage with and without an Avere vFXT cluster
2020

2121
## High-performance computing
2222

23-
* [Best Practices for Improving Azure Virtual Machine (VM) Boot Time](https://github.com/Azure/Avere/blob/master/docs/azure_vm_provision_best_practices.md) - A reproducible test of using Avere vFXT for Azure to improve boot times when spinning up thousands of compute clients quickly
23+
* [Best practices for improving Azure Virtual Machine (VM) boot time](https://github.com/Azure/Avere/blob/master/docs/azure_vm_provision_best_practices.md) - A reproducible test that uses Avere vFXT for Azure to improve boot times when spinning up thousands of compute clients quickly
2424

2525
## vFXT performance
2626

articles/avere-vfxt/avere-vfxt-deploy-overview.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -4,7 +4,7 @@ description: Overview of deploying Avere vFXT for Azure
44
author: ekpgh
55
ms.service: avere-vfxt
66
ms.topic: conceptual
7-
ms.date: 02/20/2019
7+
ms.date: 12/20/2019
88
ms.author: rohogue
99
---
1010

articles/avere-vfxt/avere-vfxt-faq.md

Lines changed: 15 additions & 16 deletions
Original file line numberDiff line numberDiff line change
@@ -4,7 +4,7 @@ description: Frequently asked questions about Avere vFXT for Azure
44
author: ekpgh
55
ms.service: avere-vfxt
66
ms.topic: conceptual
7-
ms.date: 11/06/2019
7+
ms.date: 12/19/2019
88
ms.author: rohogue
99
---
1010

@@ -20,11 +20,11 @@ Avere vFXT for Azure is a high-performance file system that caches active data i
2020

2121
### Is Avere vFXT a storage solution?
2222

23-
No. Avere vFXT is a file-system *cache* that attaches to storage environments, such as your EMC or NetApp NAS or an Azure blob container. Avere vFXT streamlines data requests from clients, and it caches the data that it serves to improve performance at scale and over time. Avere vFXT itself does not store data. It has no information about the amount of data stored behind it.
23+
No. Avere vFXT for Azure is a file-system *cache* that attaches to storage environments, such as your EMC or NetApp NAS, or an Azure blob container. Avere vFXT streamlines data requests from clients, and it caches the data that it serves to improve performance at scale and over time. Avere vFXT itself does not store data. It has no information about the amount of data stored behind it.
2424

2525
### Is Avere vFXT a tiering solution?
2626

27-
Avere vFXT does not automatically tier data between hot and cool tiers.
27+
Avere vFXT for Azure does not automatically tier data between hot and cool tiers.
2828

2929
### How do I know if an environment is right for Avere vFXT?
3030

@@ -42,9 +42,9 @@ Consider Avere vFXT for Azure if you have a file-based analytic pipeline that ru
4242

4343
* Your HPC application is based on NFSv3 clients. (In some circumstances, it can use SMB 2.1 clients, but performance is limited.)
4444

45-
The following diagram simplifies the answer to this question. The closer your workflow is to the upper right, the more likely it is that the Avere caching solution is right for your environment.
45+
The following diagram can help you answer to this question. The closer your workflow is to the upper right, the more likely it is that the Avere vFXT for Azure caching solution is right for your environment.
4646

47-
![Diagram showing that read-heavy loads with thousands of clients are better suited for Avere vFXT](media/avere-vfxt-fit-assessment.png)
47+
![Graph diagram showing that read-heavy loads with thousands of clients are better suited for Avere vFXT](media/avere-vfxt-fit-assessment.png)
4848

4949
### At what scale of clients does the Avere vFXT solution make the most sense?
5050

@@ -54,15 +54,15 @@ Typical Avere vFXT customers run demanding workloads starting at about 1,000 CPU
5454

5555
### How much data can an Avere vFXT environment store?
5656

57-
Avere vFXT is a cache. It doesn't specifically store data. It uses a combination of RAM and SSDs to store the cached data. The data is permanently stored on a back-end storage system (for example, a NetApp NAS system or a blob container). The Avere vFXT system does not have information about the amount of data stored behind it. Avere vFXT only caches the subset of that data that clients request.
57+
Avere vFXT for Azure is a cache. It doesn't specifically store data. It uses a combination of RAM and SSDs to store the cached data. The data is permanently stored on a back-end storage system (for example, a NetApp NAS system or a blob container). The Avere vFXT system does not have information about the amount of data stored behind it. Avere vFXT only caches the subset of that data that clients request.
5858

5959
### What regions are supported?
6060

6161
Avere vFXT for Azure is supported in all regions except for sovereign regions (China, Germany). Make sure that the region you want to use can support the large quantity of compute cores and the VM instances needed to create the Avere vFXT cluster.
6262

6363
### How do I get help with Avere vFXT?
6464

65-
A specialized support group offers help with Avere vFXT for Azure. Follow the instructions in [Get help with your system](avere-vfxt-open-ticket.md#open-a-support-ticket-for-your-avere-vfxt) to open a support ticket from the Azure portal.
65+
A specialized group of support staff offers help with Avere vFXT for Azure. Follow the instructions in [Get help with your system](avere-vfxt-open-ticket.md#open-a-support-ticket-for-your-avere-vfxt) to open a support ticket from the Azure portal.
6666

6767
### Is Avere vFXT highly available?
6868

@@ -73,7 +73,7 @@ Yes, Avere vFXT runs exclusively as an HA solution.
7373
Yes, customers can use more than one cloud provider with the Avere vFXT cluster. It supports AWS S3 standard buckets, Google Cloud Services standard buckets, and Azure blob containers.
7474

7575
> [!NOTE]
76-
> A software fee applies to use Avere vFXT in AWS or Google Cloud, but not with Azure.
76+
> A software fee applies to use Avere vFXT with AWS or Google Cloud storage. There is no additional software fee for using Azure blob storage.
7777
7878
## Technical: Compute
7979

@@ -160,18 +160,18 @@ Data is striped across the disks but is not encrypted. However, the disks themse
160160

161161
### What network is recommended?
162162

163-
If you're using on-premises storage with Avere vFXT, you should have a 1-Gbps or better network connection. If you have a small amount of data and are willing to copy data to the cloud before running jobs, VPN connectivity might be enough.
163+
If you're using on-premises storage with Avere vFXT, you should have a 1-Gbps or better network connection between your storage and the cluster. If you have a small amount of data and are willing to copy data to the cloud before running jobs, VPN connectivity might be sufficient.
164164

165165
> [!TIP]
166-
> The slower the network link is, the slower the initial cold reads will be. Slow reads increase the latency of the work pipeline.
166+
> The slower the network link is, the slower the initial "cold" reads will be. Slow reads increase the latency of the work pipeline.
167167
168168
### Can I run Avere vFXT in a different virtual network than my compute cluster?
169169

170170
Yes, you can create your Avere vFXT system in a different virtual network. Read [Plan your Avere vFXT system](avere-vfxt-deploy-plan.md) for details.
171171

172172
### Does Avere vFXT require its own subnet?
173173

174-
Yes. Avere vFXT runs strictly as a high availability (HA) cluster and requires multiple IP addresses to operate. If the cluster is in its own subnet, you avoid the risk of IP address conflicts, which can cause problems for installation and normal operation. The cluster's subnet can be within the existing virtual network as long as no IP addresses overlap.
174+
Yes. Avere vFXT runs strictly as a high availability (HA) cluster and requires multiple IP addresses to operate. If the cluster is in its own subnet, you avoid the risk of IP address conflicts, which can cause problems for installation and normal operation. The cluster's subnet can be within a virtual network used by other resources, as long as no IP addresses overlap.
175175

176176
### Can I run Avere vFXT on InfiniBand?
177177

@@ -219,9 +219,6 @@ In general terms, Avere vFXT for Azure supports the following systems as core fi
219219
* Dell EMC Isilon (OneFS 7.1, 7.2, 8.0, and 8.1)
220220
* NetApp ONTAP (Clustered Mode 9.4, 9.3, 9.2, 9.1P1, 8.0-8.3) and (7-Mode 7.*, 8.0-8.3)
221221

222-
> [!NOTE]
223-
> Azure NetApp Files currently is not supported.
224-
225222
* Azure blob containers (locally redundant storage only)
226223
* AWS S3 buckets
227224
* Google Cloud buckets
@@ -240,12 +237,14 @@ Support is based on the amount of demand in the field. If there are enough reven
240237

241238
### Can I use Azure Blob storage as a core filer?
242239

243-
Yes, Avere vFXT for Azure can use a block blob container as a cloud core filer.
240+
Yes, Avere vFXT for Azure can use a block blob container as a cloud core filer.
244241

245242
### What are the storage account requirements for a blob core filer?
246243

247244
Your storage account must be a general-purpose v2 (GPv2) account and configured for locally redundant storage only. Geo-redundant storage and zone-redundant storage are not supported.
248245

246+
Read [Azure Blob Storage cloud core filer](avere-vfxt-add-storage.md#azure-blob-storage-cloud-core-filer) for more details about the storage account requirements.
247+
249248
### Can I use archive blob storage?
250249

251250
No. The service-level agreement (SLA) for archive storage is not compatible with the real-time directory and file access needs of the Avere vFXT system.
@@ -254,7 +253,7 @@ No. The service-level agreement (SLA) for archive storage is not compatible with
254253

255254
Cool tier blob storage is not usually recommended for an Avere vFXT for Azure core filer. Cool tier offers lower storage costs but higher operations costs. (See [Block blob pricing](<https://azure.microsoft.com/pricing/details/storage/blobs/>) for more details.) If data will be accessed and modified or deleted frequently, please consider using the Hot tier.
256255

257-
[Access tiers](https://docs.microsoft.com/azure/storage/blobs/storage-blob-storage-tiers#cool-access-tier) gives more information about when it might make sense to use cool tier storage as a vFXT core filer.
256+
[Access tiers](https://docs.microsoft.com/azure/storage/blobs/storage-blob-storage-tiers#cool-access-tier) gives more information about when it might make sense to use Cool tier storage as a vFXT core filer.
258257

259258
### How do I encrypt the blob container?
260259

0 commit comments

Comments
 (0)